1494 / 1708
radiotherapy. Hazard ratios and 95% confidence intervals were
calculated. An interaction term was created for location and age in
order to test for effect modification between the two independent
variables. All statistical analyses were completed using SPSS
(version 15.0, Chicago, IL).
RESULTS
The characteristics of the 635 patients identified from the SEER
database are listed in Table I. Most tumors, 80.5%, were classified as
well-differentiated, or unspecified, ependymoma (9391) while the
remaining included anaplastic, papillary and myxopapillary
ependymomas. Mean age was 6.3 standard error 0.22 years and
median age 4 years. Overall, a majority of patients were male and the
predominant racewas white. Most patients (56.4%) were reported to
have received radiotherapy. The most common identified tumor
location was infratentorial.
Univariate comparisons of survival based upon age, race, gender,
and treatment appear in Table II. Radiotherapy was associated with
significantly increased survival compared with no treatment
(logrank
P
¼
0.022). There was no significant difference in survival
by gender or race, although there were trends toward improved
survival among females compared with males and whites compared
with blacks. Increasing age was associated with improved survival
(logrank
P
<
0.0001).
Tumor location was classified into spinal, supratentorial, and
infratentorial according to Figure 1. Three hundred fifty-four tumors
were included in the location analysis. Univariate analysis
demonstrated a significant difference in survival among the three
tumor locations, with improved survival among spinal tumors
(logrank
P
¼
0.001; Fig. 2). There was no difference in 5-year
survival between supratentorial and infratentorial locations. Spe-
cific 5-year survival data for the three locations appear in Table III.
Even when survival analyses for location were repeated by
including the indeterminate location cases with classification as
all supratentorial or all infratentorial, our findings were unchanged
(data not shown).
Radiotherapy for the individual tumor locations was also
compared (Table IV). Using univariate analysis, radiotherapy did
not significantly affect survival in spinal or supratentorial tumors.
Radiotherapy appeared to convey a significant survival improve-
ment for infratentorial tumors compared to no treatment in both
univariate analysis (logrank
P
¼
0.018, Fig. 3) and subgroup
multivariate analysis restricted to infratentorial tumors
(
P
¼
0.033). Using multivariate analysis for all tumor locations,
Pediatr Blood Cancer
DOI 10.1002/pbc
TABLE I. Patient Characteristics, n
¼
635
Characteristic
n
%
Gender
Male
370
58.3
Female
265
41.7
Age at initial diagnosis, years
0–4
329
51.8
5–8
104
16.4
9–12
80
12.6
13–15
67
10.5
16–18
55
8.6
Mean standard deviation
6.33 5.45
Median
4
Histologic diagnosis (ICD-0-3 code)
Ependymoma (9391)
511
80.5
Anaplastic ependymoma (9392)
119
18.7
Papillary ependymoma (9393)
1
0.2
Myxopapillary ependymoma (9394)
4
0.6
Primary tumor site
Supratentorial
106
16.7
Infratentorial
193
30.4
Spinal
55
8.7
Other
281
44.3
Race
White
510
80.3
Black
77
12.1
Other
48
7.5
Treatment
Radiotherapy
358
56.4
No radiotherapy
254
40.0
Unknown
23
3.6
n, number of children.
TABLE II. Univariate Comparison of Survival by Gender, Race, and Treatment
n
Median survival
(months)
5-year survival (%)
SE (%)
P
-value*
Gender
0.97
Male
370
97.0
56.4
3.0
Female
265
154
54.9
3.6
Race
0.20
White
510
154
56.7
2.5
Black
77
53.0
45.1
6.7
Treatment
0.022
Radiotherapy
358
154
56.8
3.0
No radiotherapy
254
91.0
53.6
3.6
Age
0–4
329
43.0
45.5
3.1
<
0.0001
5–8
104
145
56.5
6.1
9–12
80
268
63.3
6.5
13–15
67
—
75.8
6.1
16–18
55
—
77.0
7.0
n, number of children; SE, standard error. *Logrank Test.
66
McGuire et al.