paediatrics Brussels 17

1496 / 1708

radiation only for infratentorial tumors, while radiotherapy was not
associated with a difference in survival for spinal and supratentorial
tumors. While our data are retrospective and treatments were not
assigned randomly, radiotherapy appears beneficial for infratento-
rial ependymoma, counter to some prior speculation [14].
Recent work by Taylor et al. [18] suggests that specific
progenitor cells may contribute to the formation of the distinct
tumor types occurring in the supratentorial, infratentorial and spinal
regions. Our study found spinal tumors were associated with a
significantly better prognosis than both supratentorial and infra-
tentorial tumors, but no difference was observed when comparing
supratentorial to infratentorial tumors. Similar findings suggesting
no difference in survival among supratentorial and infratentorial
locations were shown in a recent study by Shu et al. [8] Together
these findings suggest that spinal tumors may represent a distinct
biological entity, but are not particularly supportive of infratentorial
and supratentorial tumors as entirely distinct.
One limitation of our study was the inability to specify location
for tumors identified as ventricle, overlapping brain, brain not
otherwise specified, overlapping, or not otherwise specified (715,
718, 719, 728, and 729). When analyzing how location affects
survival, these specific cases were eliminated because of the
inability to assign them to supratentorial, infratentorial, or spinal. In
order to ensure that there was no systematic error associated with
this determination, univariate location analysis was repeated with
all the indeterminate cases included and classified as infratentorial.
The result was compared to a second univariate analysis completed
with all the indeterminate cases classified as supratentorial. Because
there was no statistically significant difference in the two results, the
lack of site classification was judged to be random and made
exclusion of these cases from any analysis examining location
reasonable.
In a retrospective, observational study, it is always possible that
other confounding variables were not incorporated into the analysis
and influenced the results. Potential prognostic factors such as stage,
extent of surgical resection, histologic grade, and specific details of
all treatments are not consistently available over three decades in the
SEER database, and therefore could not be included in our
multivariate analysis. Specifically, histologic grade was not
analyzed in our study because there is wide variation among
institutions in tumor grading and most registrars encode ependy-
momas as ‘‘ependymomas,’’ regardless of whether well-differ-
entiated or anaplastic. Finally, advances in medicine, including new
surgical technologies, evolution of computed tomography and then
magnetic resonance imaging as well as revision of pathology
classification schema, have occurred during the time period
examined and may lead to reporting bias.
Despite these limitations, our study demonstrates how survival
varies by age, location, and radiotherapy. Understanding these
trends may help us further understand the biology and guide
refinements in the treatment of ependymoma. Distinct knowledge of
how radiation and tumor location relate to survival may guide
clinical management of ependymoma in these populations, perhaps
leading to modification of treatment guidelines for young children.
ACKNOWLEDGMENT
Courtney McGuire was supported by the Stanford University
Medical Scholars Program.
REFERENCES
1. Kricheff II, Becker M, Scheneck SA, et al. Intracranial ependy-
momas: Factors influencing prognosis. J Neurosurg 1964;21:7–14.
2. Farwell JR, Dohrmann GJ, Flannery JT. Central nervous system
tumors in children. Cancer 1977;40:3123–3132.
3. Dohrmann GJ, Farwell JR. Ependymal neoplasms in children.
Trans Am Neurol Assoc 1976;101:125–129.
4. Ries LAG, Melbert D, Krapcho M, et al. editors. SEER Cancer
Statistics Review, 1975–2004. Bethesda, MD: National Cancer
Institute 2007. based on
November 2006 SEER data submission, posted to the SEER web
site.
5. Central Brain Tumor Registry of the United States data, 1998–
2002. URL
6. Gurney JG, Smith MA, Bunin GR. CNS and miscellaneous
intracranial and intraspinal neoplasms. Ries LAG, Smith MA,
Gurney JG, et al. editors. Cancer Incidence and Survival Among
Pediatr Blood Cancer DOI 10.1002/pbc
TABLE IV. Univariate Comparison of Survival by Radiotherapy for Each Primary Tumor Site, n ¼ 339
Radiation No radiation
P-value* n Median (mo)
5-year survival
(%) SE (%) n Median (mo)
5-year survival
(%) SE (%)
Supratentorial 65 120.0 54.0 7.0 38 190.0 68.0 8.6 0.95
Infratentorial 116 116.0 57.1 5.2 68 43.0 48.2 7.1 0.018
Spinal 20 — 95.0 4.9 32 — 80.0 8.4 0.82
n, number of children; mo, months, SE, standard error. *Logrank Test.
400.00 300.00 200.00 100.00 0.00
Total survival time (months)
0.0
0.2
0.4
0.6
0.8
1.0
Cumulative survival
XRT
No XRT
Fig. 3. Comparison of survival by radiotherapy for infratentorial
tumors. XRT ¼radiotherapy. Total number of patients ¼ 184. Number
of patients in each arm: XRT¼ 116, no XRT ¼ 68. Logrank Test,
P¼ 0.018.
68 McGuire et al.