1574 / 1708
with survival. More extensive tumor resection was associ-
ated with a non-significant trend favoring survival. The
5-year survival rates were 94% for patients having gross-
total resections, 76% for those having subtotal resections,
and 67% for those having only biopsies (
p
5
0.2).
Univariate analysis (log-rank test) revealed that the fol-
lowing factors were associated with survival: tumor grade,
location, and histologic type. Patients with low-grade tu-
mors had a 5-year survival rate of 87% as compared to 27%
for those with high-grade tumors (
p
,
0.0001) (Fig. 5).
Patients with tumors of the spine had a 5-year survival rate
of 97% as compared to 68% for those with lesions of the
infratentorial brain and 62% for those with lesions of the
supratentorial brain (
p
5
0.03) (Fig. 6). Patients with the
myxopapillary subtype had a 5-year survival rate of 100%
as compared to 76% for patients with other histological
subtypes of ependymoma (
p
5
0.02) (Figs. 7 and 8). Mul-
tivariate analysis revealed that survival was independently
associated with both grade (
p
5
0.0007) and histological
subtype (
p
5
0.02), but not tumor location (
p
5
0.07).
DISCUSSION
The present analysis was performed to define the long-
term outcome of patients with ependymomas. The ideal
preoperative staging work-up should include a careful his-
tory and physical examination, magnetic resonance imaging
(MRI) of the clinically involved CNS site, and a CSF
cytology. Clinically uninvolved CNS sites should also be
imaged with MRI because of the risk of meningeal seeding.
In our series, 6% of patients were found to have CNS
seeding prior to therapy and 16% following treatment.
Prognostic factors associated with survival on univariate
analysis included tumor grade, location, and histologic type
(myxopapillary vs. other subtypes). Both tumor grade and
histologic type were associated with survival on multivari-
ate analysis. The importance of tumor grade as a determi-
nant of survival has been observed by other investigators
(7–13). Tumor grade is the most consistently reported prog-
nostic factor in the literature.
Post-operative radiotherapy resulted in a high (87%)
5-year survival rate for those with low-grade tumors. Our
data suggests that survival is improved when there is resec-
tion of as much tumor as is safely possible. For patients
having residual disease detected intraoperatively or with a
postoperative MRI scan, moderate dose radiotherapy to the
tumor bed is indicated. Garrett and Simpson reported a
dose-response for patients with ependymomas. Of their
patients who received
#
45 Gy, 5 of 18 (28%) were alive at
Table 4. The effect of tumor grade on patterns of failure
Low-grade (1–2) tumors (
n
5
70)
Patterns of failure
Local failure
Local control
Total
Distant failure
3 (4%)
4 (6%)
7 (10%)
Distant control
16 (23%)
47 (67%)
63 (90%)
Total
19 (27%)
51 (73%)
70 (100%)
High-grade (3–4) tumors (
n
5
10)
Patterns of failure
Local failure
Local control
Total
Distant failure
3 (30%)
0
3 (30%)
Distant control
6 (60%)
1 (10%)
7 (70%)
Total
9 (90%)
1 (10%)
10 (100%)
All patients (
n
5
80)
Patterns of failure
Local failure
Local control
Total
Distant failure
6 (7.5%)
4 (5%)
10 (12.5%)
Distant control
22 (27.5%)
48 (60%)
70 (87.5%)
Total
28 (35%)
52 (65%)
80 (100%)
Patients are categorized into various groups by tumor grade and the subsequent pattern of failure, including both local failures and distant
(leptomeningeal) failures.
Fig. 4. Overall survival for the entire group of patients.
Fig. 5. Overall survival by tumor grade (1 and 2 vs. 3 and 4).
956
I. J. Radiation Oncology
●
Biology
●
Physics
Volume 42, Number 5, 1998