Received: 1 June 2020
Revised: 13 July 2020
Accepted: 14 July 2020
Pediatric
Blood &
Cancer
DOI: 10.1002/pbc.28622
O N C O LO G Y: R E S E A R C H A RT I C L E
The American Society of
Pediatric Hematology/Oncology
Surveillance imaging in pediatric ependymoma
Darren Klawinski1,2
Daniel J. Indelicato3
Jobayer Hossain4
Eric Sandler1
1
Division of Pediatric Hematology/Oncology,
Nemours Children’s Specialty Care and
Wolfson Children’s Hospital, Jacksonville,
Florida
2
Department of Pediatrics, University of
Florida, Jacksonville, Florida
3
Department of Radiation Oncology,
University of Florida, Jacksonville, Florida
4
Department of Statistics and Bioinformatics,
A.I. DuPont Hospital for Children, Wilmington,
Delaware
Correspondence
Darren Klawinski, Division of Pediatric Hematology/Oncology, Nemours Children’s Specialty
Care and Wolfson Children’s Hospital, 807
Children’s Way, Jacksonville, FL 32207.
Email: Darren.Klawinski@gmail.com
The abstract for this manuscript was submitted
for presentation at the 2020 American Society
of Pediatric Hematology/Oncology conference
in Fort Worth, TX, May 6–9 (canceled due to
COVID-19) and was published in Pediatric Blood
and Cancer.
Abstract
Background: Management of pediatric patients with ependymoma includes posttreatment surveillance imaging to identify asymptomatic recurrences. However, it is unclear
whether early detection translates into improved survival. The objective was to determine whether detection of ependymoma relapses on surveillance imaging translates
into a survival benefit.
Procedure: Patients with ependymoma aged <21 years at diagnosis treated in the
Nemours’ Children’s Health System between January 2003 and October 2016 underwent chart review. Relapsed patients’ charts were assessed for details of initial therapy, surveillance imaging regimen, details of relapse including detection and therapy,
and outcome. Median follow up of the entire cohort was 6.5 years from diagnosis and
3.5 years from relapse.
Results: Ninety of 198 (45%) patients experienced relapse with 61 (68%) detected by
surveillance imaging and 29 (32%) detected based on symptoms. Five-year OS in the
surveillance group was 67% (confidence interval [CI] 55–82%, SE 0.1) versus 51% (CI
35–73%, SE 0.19) in the symptoms group (P = .073). From relapse, the 3-year OS in the
surveillance group was 62% (CI 50–78%, SE 0.11) versus 55% (CI 39–76%, SE 0.17) in
the symptoms group (P = .063) and the 3-year SPFS was 45% (CI 33–61%, SE 0.16)
in the surveillance group versus 32% (CI 19–55%, SE 0.27) in the symptoms group
(P = .028).
Conclusion: Surveillance imaging may identify recurrences in patients when they are
more amenable to salvage therapy, resulting in superior 3-year SPFS, but given limited salvage options for children with recurrent ependymoma, the survival advantage
of frequent surveillance imaging in asymptomatic patients remains ambiguous.
KEYWORDS
detection, ependymoma, prognosis, relapse, surveillance, survival
1
INTRODUCTION
a high incidence of local tumor recurrence compared to other childhood brain tumors with approximately one-third of patients experienc-
Ependymomas account for about 5% of pediatric central nervous system (CNS) tumors and are the second most common high-grade CNS
tumor in
children.1
Despite aggressive treatment, it is associated with
ing relapse.2-6
In the management of pediatric ependymoma, an essential component is serial follow-up imaging with magnetic resonance imaging
(MRI) or computed tomography (CT) with the presumption that the
early identification of asymptomatic recurrences leads to improve-
Abbreviations: CI, confidence interval; CNS, central nervous system; CT, computed
tomography; MRI, magnetic resonance imaging; OS, overall survival; SPFS, second
progression-free survival
Pediatr Blood Cancer. 2020;e28622.
https://doi.org/10.1002/pbc.28622
ments in outcome. Surveillance protocols have been proposed based
on the biological characteristics of a specific tumor and take into
wileyonlinelibrary.com/journal/pbc
© 2020 Wiley Periodicals LLC
1 of 8
2 of 8
KLAWINSKI ET AL .
account the rate of tumor growth as well as incidence of local and
metastatic
recurrence.7,8
electronic medical record. Included patients were aged 21 years and
While surveillance imaging may identify
under and were newly diagnosed with a primary ependymoma at least
tumor recurrence before symptoms develop, it is less clear that early
2 years prior to the beginning of the study. They underwent and com-
detection leads to improved outcomes and/or changes in treatment
pleted treatment that consisted of surgical resection, radiation ther-
strategies.3,7,8
apy, chemotherapy, or a combination thereof, with curative intent. At
Additionally, for many children, the practice of surveillance imag-
the end of treatment, they began a period of surveillance imaging
ing is not without problems. Performing surveillance imaging has been
with MRI to assess for disease recurrence. The schedule of surveil-
associated with false positive results that not only add cost to both fam-
lance imaging was based on the treating physician following the pat-
ilies and the healthcare system but also lead to further imaging and
tern of every 3–4 months for the first 3 years following treatment,
unnecessary further therapy. False negative results may delay neces-
every 6 months until 5 years off treatment, and yearly thereafter.
sary treatment changes, add to costs, and compromise patient quality
Surveillance images included imaging of both the brain and the spine.
and length of life.8,9 Both patients and families often experience signif-
However, spinal imaging was obtained less frequently for patients with
icant anxiety around the time of surveillance imaging.8,10 Sedation or
primary disease in the brain, with an approximately 2:1 ratio of brain
general anesthesia is often required, in particular for younger children,
to spinal images. While all images for every patient were not avail-
to obtain surveillance imaging. Wu and colleagues recently reviewed
able, imaging was available immediately prior to and at the time of
and discussed the relationship between general anesthetics and neu-
documented relapse on imaging. All patients had documented disease
rotoxicity and highlight how the developing brain may be particularly
recurrence with MRI. The majority of patients were not enrolled on
vulnerable to anesthesia leading to a risk for long-lasting impairments
treatment clinical trials and received treatment based on the exper-
of cognitive function.11
tise of the treating physician. However, more than 90% of patients were
Advancements in pediatric neuroimaging have led to MRI replacing
enrolled on the University of Florida Proton Therapy Institute outcome
CT as studies have shown that MRI has improved image quality, bet-
tracking study. Diagnosis was based on institutional pathology for the
ter differentiation between normal brain and tumor using contrast, and
majority of patients. Median follow up in the relapsed cohort was
has been better able to delineate the extent of tumor and metastases
6.5 years from diagnosis and 3.5 years from relapse. Patients with dis-
throughout the brain and spine.8,12-14 While MRI has the advantages
ease progression prior to the end of treatment or who did not undergo
of being the better imaging modality for brain tumors and involves no
initial treatment were excluded. All patients had relapse documented
ionizing radiation exposure, recent studies have shown that gadolin-
by MRI.
ium contrast deposits in various parts of brain. The significance of this
has not been determined, but it has the potential to result in long-term
neurotoxicity.15,16
2.2
Study aims
For these reasons, obtaining a better understanding of the prognostic significance of surveillance imaging is necessary as clinicians weigh
The study aims to assess the significance of current neuroimaging prac-
the positive and negative aspects for patients. While prior studies have
tices as it relates to overall prognosis. It is hypothesized that frequent
been performed to assess the prognostic value of surveillance imag-
surveillance scanning will allow earlier detection of disease recurrence,
ing in the detection of relapse in children with various brain tumors,
but will not confer benefit to OS in these high-grade brain tumors.
including ependymoma, there is a relatively small amount of literature
that includes little contemporary data addressing this question.3,7,9 In
addition, imaging techniques in previous studies included both CT and
2.3
Outcomes
MRI and were more likely to be CT if the patient presented earlier in
the study period and more likely to be MRI later in the study period.
Charts of relapsed patients were further examined to assess details of
Both methods detected relapse, but the studies did not compare the
initial therapy, surveillance imaging regimen, timing of relapse, means
two.3,7 Based on this previous literature, it is hypothesized that fre-
of detection, relapsed therapy, and outcome. The primary aim of the
quent surveillance imaging will allow earlier detection of disease recur-
study was to assess correlation between method of relapse detection
rence, but will not confer an overall survival (OS) benefit.
and survival.
2
2.4
2.1
METHODS
Study design and participants
Statistical analysis
Survival analyses were computed for each group using the KaplanMeier method. Kaplan-Meier curves were constructed to show sur-
Under an Institutional Review Board approved protocol, patients with
vival over time. OS was determined as length of time of survival from
ependymoma who were treated in the Nemours Children’s Health
diagnosis and second progression-free survival (SPFS) was determined
System in Jacksonville, Delaware, Orlando, and Pensacola between
as the length of time from relapse to disease progression. Ranges of
January 2003 and October 2016 underwent chart review via the
OS and SPFS were observed and median survival times were computed
3 of 8
KLAWINSKI ET AL .
for each cohort. Characteristics of each cohort were compared using
surveillance imaging, whereas in the surveillance group, all except one
Z-tests of proportions as well as T-tests, where appropriate. Univari-
patient were on a yearly schedule, however that patient was previ-
ate and multivariate hazard ratios were calculated with Cox propor-
ously lost to follow up and at re-presentation had a surveillance scan
tional hazard model with relationship to SPFS, OS, and survival follow-
at 14 months that showed disease recurrence. All four patients with
ing recurrence.
recurrence detected on surveillance are alive at median follow up from
relapse of 20.5 months and four of five patients detected on symptoms
are alive at median follow up of 26.5 months from recurrence.
2.5
Role of funding source
Those funding the study had no role in the study design, data collection,
4
DISCUSSION
data analysis, data interpretation, or writing of the report. DK and ES
had full access to all the data in the study while JH and DI had access to
Surveillance imaging is a practice performed in all pediatric brain
part of the data in the study. DK and ES had final responsibility for the
tumors and many other solid tumors to detect asymptomatic disease
decision to submit for publication.
recurrence with the hopes that early identification will lead to improvement in outcome with salvage therapy. In pediatric ependymoma,
this question has not been definitively answered.3,7 In this study, the
3
RESULTS
OS from initial diagnosis and from time of relapse was different for
those with asymptomatic detection on surveillance than those with
Ninety of 198 (45%) patients experienced disease progression. Of
symptomatic detection, but this failed to show statistical significance.
those patients, 61 (68%) were detected with surveillance imaging while
However, there was a significantly better 3-year SPFS observed in
29 (32%) were detected based on symptoms. Characteristics of each
asymptomatic patients with recurrences that were detected on MRI.
cohort (Table 1) were compared. Median time to relapse in both groups
In addition, univariate and multivariate analyses showed death was
was 12 months from end of treatment. Median time since last negative
almost twice as likely when recurrence was detected first by symptoms.
image in both groups was 4 months. Five-year OS in the surveillance
This difference seems largely due to our observation that patients
group was 67% (confidence interval [CI] 55–82%, SE 0.1) versus 51%
whose disease recurrence was found on surveillance imaging were
(CI 35–73%, SE 0.19) in the symptoms group (P = .073) (Figure 1). The
more likely to undergo salvage therapy than their symptomatic coun-
3-year survival from time of relapse in the surveillance group was 62%
terparts, 58 of 61 (95%) when asymptomatic compared to 24 of 29
(CI 50–78%, SE 0.11) versus 55% (CI 39–76%, SE 0.17) in the symptoms
(83%) when symptomatic. Additionally, we found that when relapse
group (P = .063) (Figure 1). Following treatment of recurrence, the
was detected, patients in the symptoms cohort were more likely to
3-year SPFS was 45% (CI 33–61%, SE 0.16) in the surveillance group
have metastatic disease 12 of 29 (41%). Those with metastatic recur-
versus 32% (CI 19 55%, SE 0.27) in the symptoms group (P = .028)
rence, no matter which cohort, were also less likely to receive sal-
(Figure 1). Hazard ratios were calculated for symptomatic detection
vage chemotherapy. Symptomatic recurrences may have represented
versus detection by surveillance imaging and in both univariate and
a greater disease burden and were felt to be less amenable to ther-
multivariate analysis were 1.83 (P = .078) and 1.93 (P = .06) for OS from
apy. Likewise, beliefs by patients and their families as well as clinicians
diagnosis, 1.86 (P = .069) and 1.88 (P = .067) for OS from relapse, and
that symptomatic disease and/or metastatic disease was more signifi-
1.84 (P = .034) and 2.12 (P = .011) for SPFS (Table 2).
cant and less amenable to salvage may have dictated a strictly palliative
In all patients, 57 of 90 (63%) experienced a local recurrence. In
approach to further management.
the surveillance cohort, 40 of 61 (66%) experienced a local recurrence,
While recurrent ependymoma was once considered fatal, there
while 21 of 61 (34%) experienced a metastatic recurrence, whereas in
are increasing options for salvage therapy including craniospinal re-
the symptoms cohort, 17 of 29 (59%) were local recurrence and 12
irradiation (if focal irradiation performed as part of initial therapy),
of 29 (41%) were metastatic recurrence. Univariate and multivariate
chemotherapy, and combination therapy.2,5,17,18 Additionally, genomic
hazard ratios for metastatic recurrence were 3.06 (P = .001) and 3.33
analysis in ependymoma has led to identification of numerous potential
(P = .001) for OS from diagnosis, 3.38 (P = <.001) and 3.21 (P = .001) for
targets that could be incorporated into salvage therapy.19 These and
OS from recurrence, and 2.10 (P = .011) and 2.17 (P = .008) for SPFS.
other salvage treatment approaches were not available or considered
Following recurrence, 82 of 90 (91%) underwent salvage therapy: 58
at the time of previous studies addressing this topic.
of 61 (95%) when detected by surveillance versus 24 of 29 (83%) when
One challenge when comparing asymptomatic and symptomatic dis-
detected by symptoms. In all patients, 55 of 57 (97%) with local recur-
eases is lead-lead time bias; the bias that asymptomatic detection
rence underwent salvage therapy and 27 of 33 (82%) with metastatic
occurs earlier in a disease course leading to the appearance of longer
recurrence underwent salvage therapy (P = .0181).
survival relative to those who present later in the course and have
Beyond 5 years from initial therapy, there were four of 61 (7%)
symptomatic detection. In our study, it was found that the median time
recurrences detected based on surveillance, while five of 29 (17%) in
to relapse and the median time since last negative surveillance image
the symptoms group occurred after 5 years (overall 10% occur after
in both cohorts did not differ. This suggests that lead-time bias was
5 years). In the symptoms group, two patients were not getting yearly
minimal in this study. It would therefore be reasonable to assume that
4 of 8
TA B L E 1
KLAWINSKI ET AL .
Comparison of relapsed cohorts
Symptoms (n = 29)
Surveillance (n = 61)
Age at diagnosis (mos)
P value (95% CI)
.0369
Range
8-226
5-168
Median
36
28
Gender
.0489
Male (n = 58)
22 (76%)
36 (59%)
Female (n = 32)
7 (24%)
25 (41%)
Histology
Grade I
0 (0%)
2 (3%)
.1731
Grade II
9 (31%)
20 (33%)
.8509
Grade III
20 (69%)
39 (64%)
.6407
17 (59%)
47 (77%)
.0945
Location at diagnosis
Infratentorial
Supratentorial
12 (41%)
12 (20%)
.0483
Spine
0 (0%)
2 (3%)
.1731
Brain
1 (3%)
0
.176
Spinal cord
0
1 (2%)
.1342
Gross total
18 (62%)
42 (69%)
.5225
Near total
3 (10%)
8 (13%)
.6744
Subtotal resection
8 (28%)
11 (18%)
.3089
Metastatic at diagnosis
Resection
Tx post resection
Observation
4 (14%)
5 (8%)
.4196
RT only
13 (45%)
25 (41%)
.7244
Chemotherapy only
3 (10%)
19 (31%)
.0108
RT + chemotherapy
9 (31%)
12 (20%)
.2786
Time to relapse (mos)
.1198
Range
0-120
0-78
Median
12
12
Last negative image to relapse detection (mos)
.0908
Range
2-67
2-14
Median
4
4
Local
17 (59%)
40 (66%)
Metastatic
12 (41%)
21 (34%)
None
5 (17%)
3 (5%)
.1161
Relapse
.5293
Tx post relapse
Surgery only
0 (0%)
2 (3%)
.1731
Radiation only
1 (3%)
2 (3%)
1
Chemotherapy only
2 (7%)
2 (3%)
.4505
RT + chemotherapy
1 (3%)
2 (3%)
1
Surgery + RT
9 (31.0%)
29 (48%)
.1176
Surgery + chemotherapy
3 (10%)
7 (12%)
.7767
Surgery + RT + chemotherapy
8 (28%)
14 (23%)
.6197
Abbreviations: Mos, months; RT, radiation therapy; Tx, treatment.
KLAWINSKI ET AL .
5 of 8
F I G U R E 1 Kaplan-Meier Curves for survival probability based on relapse detection by modality surveillance (solid) and symptoms (dashed).
(A) Overall survival from diagnosis, (B) overall survival from relapse, and (C) second progression-free survival
6 of 8
KLAWINSKI ET AL .
TA B L E 2
Hazard ratios
Univariate analysis
Second progression-free survival
Overall survival from diagnosis
Overall survival from relapse
HR
HR
HR
95% CI
P-value
95% CI
P-value
95% CI
P-value
Age at diagnosis (mos)
0.99
0.98-1.01
.273
0.98
0.96-1.00
.02
0.99
0.97-1.01
.255
Male gender
1.47
0.81-2.65
.206
1.89
0.89-4.01
.1
1.86
0.87-3.96
.11
Symptomatic detection
1.84
1.05-3.24
.034
1.83
0.93-3.60
.078
1.86
0.95-3.62
.069
Grade III histology
1.06
0.59-1.88
.856
1.16
0.58-2.31
.684
1.30
0.64-2.67
.467
Supratentorial disease
0.55
0.28-1.08
.083
0.68
0.31-1.5
.341
0.63
0.28-1.39
.25
NTR
1.08
0.45-2.58
.863
1.75
0.70-4.40
.235
1.85
0.73-4.69
.192
STR
1.66
0.85-3.71
.136
2.11
1.00-4.45
.05
2.25
1.06-4.76
.035
Metastatic relapse
2.10
1.19-3.71
.011
3.06
1.55-6.05
.001
3.38
1.71-6.67
<.001
0.99
0.97-1.00
.071
0.97
0.95-0.99
.006
0.99
0.97-1.01
.147
Multivariate analysis
Age at diagnosis (mos)
Symptomatic detection
2.12
1.19-3.80
.011
1.93
0.97-3.83
.06
1.88
0.96-3.7
.067
Metastatic relapse
2.17
1.23-3.84
.008
3.33
1.66-6.67
.001
3.21
1.63-6.34
.001
Abbreviations: HR, hazard ratio; mos, months; NTR, near total resection; STR, subtotal resection.
symptomatic recurrences were similar to asymptomatic recurrences,
DATA AVAILABILITY STATEMENT
but simply in a location that was more likely to result in symptoms.
The data that support the findings of this study are available on request
One limitation of this study was that molecular testing was not available for the majority of the patients and therefore those factors could
from the corresponding author. The data are not publicly available due
to privacy or ethical restrictions.
not be analyzed regarding prognosis and risk of relapse in relation to
surveillance imaging. If a specific cohort defined molecularly showed
CONFLICT OF INTEREST
much poorer prognosis, a difference may or may not be seen in OS or
The authors declare that there is no conflict of interest.
SPFS between asymptomatic and symptomatic recurrences and further research is necessary to address this question.
Data from this cohort show that recurrence detected on surveillance was associated with a higher probability of survival and more
ORCID
Darren Klawinski
Daniel J. Indelicato
https://orcid.org/0000-0003-4524-2743
https://orcid.org/0000-0001-5765-1873
often lead to salvage therapy and improved SPFS from relapse. It
remains unclear if this leads to a significant difference in OS, but the
associated benefits should be individualized to patients and families
and weighed against any costs or risks. Based on the median time
to relapse of 12 months from end of treatment and few experiencing relapse after 5 years, an appropriate surveillance imaging schedule would be more frequent imaging for the first 3 years off treatment,
then spacing of imaging to every 6 months until 5 years off treatment,
followed by yearly imaging afterward. These data therefore underscore current surveillance imaging recommendations.
5
CONCLUSION
Given the current limited salvage options for children with recurrent
ependymoma, the survival advantage of frequent surveillance imaging in asymptomatic patients remains ambiguous. Surveillance imaging may identify recurrences in patients when they are more amenable
to salvage therapy, resulting in superior 3-year progression-free survival. However, further research with larger cohorts is necessary to
fully define the role of surveillance imaging.
REFERENCES
1. Ostrom QT, Gittleman H, Truitt G et al. CBTRUS statistical report:
primary brain and other central nervous system tumors diagnosed in
the United States in 2011–2015. Central Brain Tumor Registry of the
United States (CBTRUS). Neuro Oncol. 2018;20(S4):1-86.
2. Indelicato DJ, Bradley JA, Rotondo RL, et al. Outcomes following proton therapy for pediatric ependymoma. Acta Oncol. 2018;57(5):644648.
3. Good CD, Wade AM, Hayward RD, et al. Surveillance neuroimaging in
childhood intracranial ependymoma: how effective, how often, and for
how long? J Neurosurg. 2001;94:27-32.
4. Merchant TE, Li C, Xiong X, Kun LE, Boop FA, Sanford RA. Conformal
radiotherapy after surgery for pediatric ependymoma: a prospective
study. Lancet Oncol. 2009;10:258-266.
5. Merchant TE. Current clinical challenges in childhood ependymoma: a
focused review. J Clin Oncol. 2017;35:2364-2369.
6. Massimo M, Miceli R, Giangaspero F, et al. Final results of the second
prospective AIEOP protocol for pediatric intracranial ependymoma.
Neuro Oncol. 2016;18:1451-1460.
7. Steinbok P, Hentschel BS, Cochrane DD, Kestle JR. Value of postoperative surveillance imaging in the management of children with some
common brain tumors. J Neurosurg. 1996;84:726-732.
8. Weiser DA, Kaste SC, Siegel MJ, Adamson PC. Imaging in Childhood
cancer: a Society for Pediatric Radiology and Children’s Oncology
7 of 8
KLAWINSKI ET AL .
9.
10.
11.
12.
13.
14.
15.
Group Joint Task Force Report. Pediatr Blood Cancer. 2013;60:12531260.
Kovanlikaya A, Karabay N, Cakmakci H, Uysal K, Olgun N, Ergor G.
Surveillance imaging and cost effectivity in pediatric brain tumors. Eur
J Radiol. 2003;47:188-192.
Deimling GT, Bowman KF, Sterns S, Wagner LJ, Kahana B. Cancerrelated health worries and psychological distress among older adult,
long-term cancer survivors. Psychooncology. 2006;15:306-320.
Wu L, Zhao H, Weng H, Ma D. Lasting effects of general anesthetics
on the brain in the young and elderly: “mixed picture” of neurotoxicity, neuroprotection, and cognitive impairment. J Anesth. 2019;33:
321.
Lu H, Nage-Poetscher LM, Golay X, Lin D, Pomper M, van Zijl PCM.
Routine clinical brain MRI sequences for use at 3.0 Tesla. J Magn Reson
Imaging. 2005;22:13-22.
Nobauer-Huhmann IM, Ba-Ssalamah A, Mlynarik V, et al. Magnetic
resonance imaging contrast enhancement of brain tumors at 3 Tesla
versus 1.5 Tesla. Invest Radiol. 2002;37(3):114-119.
Heiss WD, Raab P, Lanfermann H. Multimodality assessment of brain
tumors and tumor recurrence. J Nucl Med. 2011;52:1585-1600.
McDonald RJ, McDonald JS, Kallmes DF, et al. Intracranial gadolinium deposition after contrast-enhanced MR imaging. Radiology.
2015;275(3):772-782.
16. Salerno S, Granata C, Trapenese M, et al. Is MRI imaging in pediatric
age totally safe? A critical reprisal. Italian Society of Medical Radiology
2018. La radiologa medica 2018
17. Tsang DS, Burghen E, Klimo P, Jr, Boop FA, Ellison DW, Merchant
TE. Outcomes after reirradiation for recurrent pediatric intracranial
ependymoma. Int J Radiat Oncol Biol Phys. 2018;100(2):507-515.
18. Tsang DS, Murray L, Ramaswamy V, et al. Carniospinal irradiation as
part of re-irradiation for children with recurrent intracranial ependymoma. Neuro Oncol. 2019;21(4):547-557.
19. Gajjar A, Bowers DC, Karajannis MA, Leary S, Witt H, Gottardo
NG. Pediatric brain tumors: innovative genomic information is
transforming the diagnostic and clinical landscape. J Clin Oncol.
2015;33(27):2986-3000.
20. Klonou A, Piperi C, Gargalionis AN, Papavassiliou AG. Molecular basis
of pediatric brain tumors. Neuromol Med. 2017;19:256-270.
How to cite this article: Klawinski D, Indelicato DJ, Hossain J,
Sandler E. Surveillance imaging in pediatric ependymoma.
Pediatr Blood Cancer. 2020;e28622.
https://doi.org/10.1002/pbc.28622