369

Pilocytic Astrocytoma: Correlation

Downloaded from www.ajronline.org by 112.215.244.130 on 09/21/17 from IP address 112.215.244.130. Copyright ARRS. For personal use only; all rights reserved

Between the Initial Imaging Features and

Clinical Aggressiveness

JohnA. strongi OBJECTIVE. Astrocytomas are classified as either fibrillary or pilocytic on the
H. Paul Hatten, Jr.2 basis of their histologic appearance. The imaging features of the fibrillary astrocy-
toma correlate closely with the tumors clinical aggressiveness and are, therefore,
MarkT. Brown3
useful in predicting prognosis. Correlation between the imaging features and the din-
J#{228}rg
F. Debatin1
ical aggressiveness of the pilocytic astrocytoma, however, Is not well established.
Henry S. Friedman4
Accordingly, we compared the initial MR and CT appearances of the lesion with tumor
w. Jerry Oakes5 aggressiveness as seen clinically to determine if a correlation exists.
Robert Ti1 MATERIALS AND METHODS. We retrospectively evaluated the Initial MR images or
CT scans of 32 consecutive patients who had a histologic diagnosis of pilocytic
astrocytoma. The lesions were evaluated with regard to location, size, calcification,
morphology, and degree of contrast enhancement. These initial imaging features
were correlated with the aggressiveness of the tumor as seen clinically. Tumors were
classified as aggressive or nonaggressive on the basis of their clinical manifesta-
tions. Patients with clinically aggressive lesions had progressive symptoms and
radiologic evidence of tumor progression or recurrence within an unusually short
period. Patients with clinically nonaggressive lesions had a more indolent course,
either improving or remaining stable, on both clinical and radiologic evaluations. In
12 patients, the tumor was classified as aggressive clinically, either progressing or
recurring within a median time of 7.5 months (range, 2.5-118 months) from the Initial
diagnosis. The remaining 20 patients had a clinically nonaggressive course.
RESULTS. In our series of patients, lesion size and location were not significantly
different between the nonaggressive and aggressive tumors, as noted clinically. Fur-
thermore, the aggressive and nonaggressive tumors were similar with regard to the
presence or absence of calcium. Most tumors in both groups showed either moder-
ate or marked enhancement and were multilobular.
CONCLUSION. The initial CT and MR features of pilocytic astrocytoma are unrell-
able for predicting which lesions will behave in a more aggressive manner clinically
and have a poor prognosis.

AJR 1993;161 :369-372

Astrocytomas are classified as either fibnillary on pilocytic on the basis of their

histologic characteristics. The imaging features of a fibnillany astrocytoma come-
Received November 12, 1992; accepted after
late strongly with both the clinical aggressiveness of the tumor and the patients
revision April 16, 1993.
1 Departmentof Radiology, Box 3808, Duke Uni- prognosis [i]. However, the correlation between the imaging features and clinical
versity Medical Center, Durham, NC 27710. Ad- aggressiveness of a pilocytic astrocytoma is not well established.
dress correspondence to J. A. Strong. Classically, pilocytic astrocytomas are associated with a relatively benign clini-
2Radiology Associates of Birmingham, 1920 cal course. This may reflect the tumors slow growth and discrete nature, which
Huntington Rd., Birmingham, AL 35209. allows complete surgical resection [2]. Some of these tumors, however, are more
3DepartmentofMedicine, Duke University Med- aggressive. Clinically, these more aggressive tumors show relentless progression
ical Center, Durham, NC 27710.
or recurrence and have a poor prognosis. Use of specific radiologic predictors to
4Department of Pediatrics, Duke University
distinguish aggressive from nonaggressive pilocytic lesions would have great
Medical Center, Durham, NC 27710.
impact on the treatment of these patients. The purpose of this study was to deter-
5Depantment of Surgery, Duke University Medi-
cal Center, Durham, NC 27710. mine if the initial preoperative CT on MR imaging findings could be used to differ-
0361-803X/93/i612-0369
entiate between those tumors that have a relatively benign clinical course and
American Roentgen Ray Society those that are more aggressive.
 370 STRONG ET AL. AJR:161 , August1993

Materials and Methods in 20 patients, the tumor was classified as clinically nonaggres-
We retrospectively evaluated the initial CT scans and MR images sive. Both clinical and radiologic evaluations showed that patients
who had clinically nonaggressive lesions had a more indolent
of 32 consecutive patients with a histologic diagnosis of pilocytic
astrocytoma who were seen at our institution in the past 20 years.
course, either improving or remaining stable. The median follow-up
Only patients in whom CT scans or MR images were obtained time for this group was 30 months (range, 1-86 months).
Downloaded from www.ajronline.org by 112.215.244.130 on 09/21/17 from IP address 112.215.244.130. Copyright ARRS. For personal use only; all rights reserved

before biopsy, radiation, chemotherapy, or surgical resection were

included. The patients were 2 months to 59 years old (mean, 14
years); 22 were female and 10 were male. Results
The images evaluated included 17 CT scans, 16 of which were In our series of 32 patients, lesions were distributed pre-
contrast enhanced. The CT scans were obtained as 5-mm contigu-
pondenantly about the hypothalamus and optic chiasm in
ous axial sections from the skull base to the vertex. Fifteen MR imag-
both the patients with aggressive tumors (four of 12) and
ing studies were done with a 1 .5-T MR system. Standard protocols
for Ti -weighted (e.g., 500/20 [TRITE]) and T2-weighted (e.g., 2200/ those with nonaggressive tumors (i 3 of 20). Four lesions
80) spin-echo (SE) pulse sequences were continually upgraded dun- were located in the cortex (two aggressive, two nonaggres-
ing the 20 years in which the images were obtained. Studies were sive) and five within the midbrain (three aggressive, two non-
generally performed with a 128 x 256 or 256 x 256 acquisition matrix, aggressive). Two nonaggressive and one aggressive lesion
a 24-cm field of view, one excitation, 5-mm-thick sections, and a 2.5- originated in the cerebellum. One aggressive lesion was
mm intenslice gap. Eleven of 1 5 examinations were performed after located in the pineal region, and one lesion from each group
lv administration of gadopentetate dimeglumine. originated in the spinal cord.
All cases were reviewed independently and in random order by Tumor size was variable and independent of clinical
two radiologists who did not know the patients clinical course. The
aggressiveness. In the nonaggressive group, tumors ranged
lesions were evaluated with regard to location, size, calcification,
from i .5 to 7.0 cm in greatest diameter. The aggressive
morphology, and degree of contrast enhancement when contrast
material had been administered. lesions ranged from 2.8 to 7.0 cm in greatest diameter. One
Tumors were classified as aggressive or nonaggressive on the spinal lesion was in each of the groups. The aggressive-
basis of their clinical manifestations. Patients with aggressive behaving lesion was more extensive, extending from the
lesions had progressive symptoms, with nadiologic evidence of lower cervical to the upper lumbar spine, whereas the
tumor progression or recurrence within an unusually short interval. benign-behaving lesion was localized to the thoracic spine.
In 12 patients with clinically aggressive tumors, progression or Almost 40% (i2/32) of the lesions contained calcifications
recurrence occurred within a median time of 7.5 months (range 2.5- that were either punctate or coarse. Six of the 12 aggressive
118 months) from the initial diagnosis. Within this group, four
lesions showed calcifications, as did six of the 20 nonag-
patients died, three who had leptomeningeal spread and one who
gressive lesions. The five lesions in the cerebellum and spi-
had skeletal metastasis. Autopsy was performed in one of the
nal cord contained no calcification.
patients who had leptomeningeal spread and showed pilocytic
tumor. Biopsy of a femoral lesion in the patient who had skeletal
The clinically aggressive and nonaggressive lesions were
involvement showed pilocytic astrocytoma. The remaining eight similar with regard to morphology. Within both subgroups,
patients required a second subtotal resection for tumor recurrence. lesions located in the optic chiasm and hypothalamus were
The pathologic diagnosis at each subsequent resection remained multilobular, with a greater than 90% solid component (Figs.
pilocytic astrocytoma. i-3). The lesions in the cortex, midbrain, and pineal regions,

Fig. 1 -AggressIve pilocytic astrocytoma in a 17-month-old boy. Lesion showed continued Fig. 2.-Nonaggressive pilocytic astrocytoma in
progression with development of leptomeningeal spread. a 17-month-old girl who had decreased visual acuity
A and B, Contrast-enhanced sagittal (A) and coronal (B) MR images (SE 500/20) show large and ataxia. Biopsy confirmed pliocytic astrocytoma.
solId lobular mass with diffuse enhancement; epicenter of mass is In hypothalamic region. En- Treatment wIth chemotherapy resulted in clinical im-
hancing metastatic lesion is In posterior part of cervical spinal cord (arrow). provement and radiologic evidence of tumor regres-
slon within 3 months. Contrast-enhanced sagfttal
MR image (SE 500/20) shows large, multilobular, in-
homogeneously enhancing mass with epicenter in
region of optIc chiasm and extension into hypothal-
amus.
 AJA:161, August 1993 PILOCYTIC ASTROCYTOMA 37i

within both subgroups, showed a more variable appearance, aggressive pineal and spinal cord lesions were performed
ranging from primarily cystic to solid. The two clinically non- without IV contrast material.
aggressive lesions in the cerebellum had the appearance of
a cystic lesion with a mural nodule (Fig. 4), whereas the din-
Discussion
ically aggressive lesion was primarily solid (Fig. 5). Both spi-
Downloaded from www.ajronline.org by 112.215.244.130 on 09/21/17 from IP address 112.215.244.130. Copyright ARRS. For personal use only; all rights reserved

nal cord lesions were primarily solid. Pilocytic astrocytomas represent a subgroup of neo-
On the images obtained after contrast material was ad- plasms that are reported to follow a more benign clinical
ministered, 23 of 27 lesions showed either moderate on course than other astrocytomas [1 , 3]. In this study, however,
marked enhancement. The nonenhancing tumors included we identified a definite subset of patients who had clinically
two of i 0 nonaggressive lesions in the region of the hypo- aggressive pilocytic astrocytomas. In the series of 32 con-
thalamus/optic chiasm and one lesion in the midbrain in both secutive cases of pilocytic astrocytoma seen at our institu-
the nonaggressive and aggressive groups. The mural nod- tion in the past 20 years, more than one thind followed an
ule of the nonaggressive cerebellan lesion was markedly unusually aggressive course, with progression or recurrent
enhanced, as was the entire solid component of the aggres- tumor identified within a median time of 7.5 months from the
sive cerebellar lesion. The nonaggressive lesion in the spinal initial diagnosis despite concurrent treatment. The reason for
cord showed marked enhancement; the examinations of the the aggressive behavior of the pilocytic astrocytomas in this

Fig. 3.-Nonaggressive pilocytic astrocytoma in a 4-year-old boy who was failing to thrive. Biopsy confirmed pliocytlc astrocytoma; patIent was still
stable 6 months after diagnosis.
A and B, Contrast-enhanced axial CT scan (A) and Ti-weighted MR Image (SE 500/20, B) show enhancing hypothalamic lesion.
C, Unenhanced axial T2-weighted MR image (SE 2200/80) shows hyperintense hypothalamic lesion.

Fig. 4.-Nonaggressive pilocytic astrocyto-

ma in an 8-year-old boy who had total tumor re-
section and was still free of tumor i5 months
after diagnosis.
A, Unenhanced sagfttal MR image (SE 500/
20) shows midline cerebellar lesion that is pri-
manly cystic but has a solid tumor component.
B, Contrast-enhanced coronal MR image (SE
500/20) shows large cystIc lesion with intense-
ly enhancing mural nodule.
 372 STRONG ET AL. AJA:161 , August 1993
Downloaded from www.ajronline.org by 112.215.244.130 on 09/21/17 from IP address 112.215.244.130. Copyright ARRS. For personal use only; all rights reserved

A B C
Fig. 5.-Aggressive pilocytic astrocytoma in a 4-year-old boy who had headache, nausea, vomiting, and ataxia. Patient had complete surgical resec-
tion of tumor, which recurred within 7 months.
A, Unenhanced axial CT scan shows diffuse low attenuation in left cerebellum.
B, Contrast-enhanced axial CT scan shows large, predominantly solid midline cerebellar lesion with inhomogeneous enhancement. Extensive pen.
tumoral edema is noted.
C, Unenhanced axial T2-weighted MR image (SE 2200/80) shows hypenintense lesion with extensive penitumoral edema.

series, compared with most other studies describing pilocytic are unreliable in predicting clinical behavior, we recommend
astrocytoma as a benign entity, is unclear. Although the regular follow-up of these patients.
selection of patients at a tertiary cane facility may have con- Although the optimal follow-up imaging schedule is
tnibuted, only two patients were referred because of failure of unknown, it is prudent to follow up these patients closely for
therapy; 30 were referred before diagnosis. at least the first year after a therapeutic intervention to
Identification of histologic, clinical, on specific nadiologic assess the behavior of the tumor. If the lesion appears to be
predictors as a means of Separating aggressive from nonag- clinically stable, the interval of follow-up scanning can then
gressive pilocytic astrocytomas would be desirable for the be gradually lengthened. In practice, it seems reasonable to
treatment of these patients. In the evaluation of fibnillary do imaging studies at least every 4 months for the first year,
astrocytic neoplasms, there is general agreement as to every 6 months for the next 2 years, and then yearly thereaf-
which histologic variables have prognostic significance [i]. ten, as long as no evidence of tumor growth is present. Obvi-
The gross and
microscopic differentiation of pilocytic astro- ously, if the tumor is progressing, on if the patients condition
cytomas with regard to clinical behavior, however, remains worsens, more frequent imaging and clinical follow-up will be
unsatisfactory [4]. The analysis of histologic variables to date required.
has shown no correlation with tumor recurrence on cure [i,
4, 5]. Correlation of the clinical features with tumor aggnes-
siveness has also had prognostic shortcomings (Brown MT REFERENCES
et al., presented at the American Academy of Neurology
1. Berger PC, Scheithauer BW, Vogel SE Surgicalpathology ofthe nervous
meeting, April i991).
system and its coverings. New York: churchill Livingstone, 1991
We evaluated the radiologic images with regard to tumor 2. Garcia DM, Fulling KH. Juvenile pilocytic astrocytoma of the cerebrum in
distribution, size, morphologic appearance, calcification, and adults. J Neurosurg 1985:63:382-386
degree of contrast enhancement. However, we were unable 3. Clark GB, Henry JM, McKeever PE. Cerebral pilocytic astrocytoma. Can-
to distinguish those tumors that would follow an aggressive cer 1985;56:11 28-1133
4. Austin EJ, Alvord EC. Recurrences of cerebellar astrocytomas: a violation
course from the tumors that would behave in a more benign of Collins law. J Neurosurg 1988;68:41-47
manner. Because a significant number of pilocytic astrocyto- 5. Shapiro K, Shulman K. Spinal cord seeding from cerebellar astrocytomas.
mas follow an aggressive course and the imaging features Childs NeivSyst 1976;2:177-1 86