University of Southern California Dissertations and Theses

Carboplatin and vincristine chemotherapy for progressive low grade gliomas in pediatric patients with or without neurofibromatosis type 1 (NF1)

(USC Thesis Other)

Carboplatin and vincristine chemotherapy for progressive low grade gliomas in pediatric patients with or without neurofibromatosis type 1 (NF1)

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content
Carboplatin and Vincristine Chemotherapy for Progressive Low
Grade Gliomas in Pediatric Patients with or without
Neurofibromatosis Type 1 (NF1)

by

Caihong Xia

A Thesis Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(APPLIED BIOSTATISTICS AND EPIDEMIOLOGY)

December 2014

Copyright 2014 Caihong Xia

ii

ACKNOWLEDGEMENTS

This work was done under the supervision and guidance of my mentor, Dr. Richard
Sposto. I would express my deepest gratitude and appreciation to Dr. Sposto for his
invaluable guidance and tremendous help to complete this thesis.

I would like to convey my great thanks and gratitude to my committee members, Dr.
Wendy Mack and Dr. Stanley Azen, for their thoughtful comments and advice on my
thesis.

I would specifically thank Dr. Joann Ater and the COG group for offering me this
precious opportunity to do data analysis for their excellent project.

Last but not least, I’d like to thank my family for their understanding and endless love
through the duration of my study. This work would not have been possible without
their support.

iii

TABLE OF CONTENTS

ACKNOWLEDGEMENTS ii

LIST OF FIGURES iv

LIST OF TABLES v

ABSTRACT vi

CHAPTER I: INTRODUCTION 1

CHAPTER II: METHODS 6

Patients 6
Study Design 7
Treatment Protocol 8
Statistical Analysis 8

CHAPTER III: RESULTS 11

Patient characteristics 11
Treatment Failures 11
Overall Outcome and Effect of NF status 12
Prognostic Factors 18
Toxicity 19

CHAPTER IV: DISCUSSION 45

REFERENCES 49

iv

LIST OF FIGURES

Fig 1: CONSORT diagram 21

Fig 2: Event-free survival (EFS) for NF1 or non-NF patients 22

Fig 3: Event-free survival by age at study entry 23

Fig 4. Event-free survival by the amount of residual tumor at study entry 24

Fig 5. Event-free survival by pathology review at study entry 25

Fig 6. Event-free survival by extent of resection 26

Fig 7. Event-free survival by tumor site 27

Fig 8. Event-free survival by tumor site (optic chiasm / hypothalamic vs others) 28

Fig 9. Event-free survival for NF1 or non-NF patients with optic pathway/
hypothalamic tumors 29

Fig 10. Event-free survival by age at study entry for patients with optic
pathway/ hypothalamic tumors 30

Fig 11. Event-free survival by amount of residual tumor at study entry
for patients with optic pathway/ hypothalamic tumors 31

Fig 12. Event-free survival by pathology review for patients with optic
pathway/ hypothalamic tumors 32

Fig 13. Event-free survival by extent of resection for patients with optic
pathway/ hypothalamic tumors 33

Fig14. The amount of residual tumor as a prognostic factor in NF1 and
non-NF patients 34

Fig15. Age as a prognostic factor in NF1 and non-NF patients 35

Fig16. Pathology review as a prognostic factor in NF1 and non-NF patients 36

Fig17. Tumor site as a prognostic factor in NF1 and non-NF patients 37

Supplemental Fig 1. Overall survival (OS) in NF1 and non-NF patients 42

Supplemental Fig 2. Overall survival (OS) for optic pathway/ hypothalamic
tumors in NF1 and non-NF patients 43

v

LIST OF TABLES

Table 1: Patient Characteristics 38

Table 2: Multivariate risk factors associated with EFS outcome 39

Table 3: Interaction between NF status and amount of residual tumor 40

Table 4: Cumulative probability of toxicity by End of Chemotherapy 41

Supplemental Table 1: Multivariate risk factors associated with OS outcome 44

vi

ABSTRACT

Background: To evaluate and compare the therapeutic effects and toxicity of
chemotherapy for low grade gliomas (LGG) in Neurofibromatosis Type 1 (NF1) and
non-NF children, patients less than 10 years of age with eligible progressive LGG or
with incomplete excision of primary LGG requiring immediate treatment were
enrolled and treated on the COG A9952 protocol with carboplatin and vincristine
(CV).

Methods: Patients with NF1 were non-randomly assigned to regimen CV. Patients
without NF were randomly assigned to CV or thioguanine, procarbazine, CCNU, and
vincristine (TPCV). Patients receiving TPCV were excluded from this analysis.
Pearson chi-square test was used to test the comparability of NF1 and non-NF patients
on regimen CV therapy in terms of their possible prognostic factors and stratification
factors. Primary end points were event-free survival (EFS) and overall survival (OS).
Events were defined as tumor progression, death from any cause, and second
malignant neoplasm (SMN). Nonparametric EFS and overall survival curves were
computed using the product-limit (Kaplan-Meier) estimates. Multivariate Cox
regression analysis was used to analyze possible prognostic factors for the risk of
recurrence. Cumulative probabilities of toxicity were obtained using life table
methods.

Results: A total of 131 eligible patients with NF1 were non-randomly assigned to CV:
43 (32.8%) NF1 patients had events and 7 (5.3%) died. Three SMNs occurred with

vii

NF1 patients receiving CV (CV-NF), at a median of 7.8 years (range 7.3 to 9.4 years)
after enrollment. The SMN was the first event in 2 patients. Median follow-up of NF1
patients without an event was 6.6 years. 129 eligible non-NF patients were randomly
assigned to CV: 81 (62.8%) experienced an event and 20 (15.5%) died. No SMNs
occurred with CV-non-NF patients. Median follow-up of non-NF patients without an
event was 6.7 years. The 5-year EFS for the CV-NF group was 69% ± 4% versus 38%
± 5% for the CV-non-NF group (P<0.001). The 5-year OS for the CV-NF group was
98% ± 1% versus 87% ± 3% for the CV-non-NF group (P=0.048). Patients at older
age (>age 3 years) with smaller amount of residual tumor had significantly better
prognosis than younger patients with larger amount of residual tumor (trend p-value
for age=0.05, trend p-value for amount of residual tumor=0.0005).

Conclusions: NF1 children tolerated CV well and had a superior event-free survival
and overall survival compared to CV-treated children without NF. They also had
decreased risk of grade 3 or 4 toxicities compared to non-NF patients.

1

CHAPTER I: INTRODUCTION

Pediatric low-grade gliomas (LGGs) are the most common childhood brain tumor.
LGGs constitute around thirty-six percent of brain tumors among children younger
than 10 years of age (Farwell, Dohrmann and Flannery, 1977). LGGs are comprised
of astrocytic tumors, oligodendroglial tumors, and mixed glial-neuronal tumors.
Pilocytic astrocytoma (WHO grade 1) and diffuse fibrillary astrocytoma (WHO grade
2) are the two most common subtypes of LGG in children. Although these tumors
vary in their clinical behavior, the majority of LGGs progress very slowly and do not
undergo malignant transformation(Sievert and Fisher, 2009). Therefore, LGGs
generally don’t require specific intervention. However, a small percentage of LGGs
progress rapidly and require immediate treatment. For patients in whom total excision
is possible, prognosis is excellent with more than 90% survival at 10 years following
surgery alone(Gajjar et al., 1997). In contrast, some patients have minimal potential
for resection, especially those whose tumor is located in a critical, unresectable
location, such as the optic pathway/hypothalamic, deep midline, and brain stem.
Although children with incompletely resected LGGs treated with subsequent
conformal radiotherapy have 74% 10-year event-free survival (EFS) rates,
practitioners want to limit radiotherapy to older children after failure of other
treatments, due to the well-described and often devastating late effects of radiotherapy
(Merchant et al., 2009). Radiotherapy may produce permanent toxicities such as
endocrinopathies, vasculopathies, optic nerve injury, secondary malignancies, and
neuropsychological deficits. Therefore, chemotherapy has taken a prominent role in
the treatment of pediatric LGGs. Present recommendations are that patients less than

2

8 years of age be treated with chemotherapy as the primary therapy to avoid long-term
cognitive and neuroendocrine side effects seen with surgery and radiation therapy in
young children (Meister, 2011).

Neurofibromatoses are genetic disorders delineated as neurofibromatosis type 1 (NF1)
and neurofibromatosis type 2 (NF2). Neurofibromatosis type 1 (NF1) is among the
most common inherited neurological disorder, affecting approximately one in every
3000 people worldwide. It affects males and females of all races equally (Arun and
Gutmann, 2004). NF2 is much less common than NF1, affecting about 1 in every
40,000 people. NF2 will not be discussed in this study. The NF1 gene is a large gene
on the long arm of chromosome 17 which encodes a protein known as neurofibromin,
a GTPase activating protein (GAP) for Ras, which is involved in the negative
regulation of Ras signaling. Many cells in the nervous system produce the
neurofibromin protein, including neurons, oligodendrocytes and Schwann cells.
Neurofibromin acts as a tumor suppressor which keeps cells from growing and
dividing too rapidly, so the cells can grow at a normal and controlled pace. Mutations
in the NF1 gene may disrupt the function of neurofibromin and compromise Ras
inactivation. The activated Ras signaling pathway will then lead to accelerated growth
and proliferation of neural crest-derived cells, and ultimately tumor formation
(Dilworth et al., 2006). As a result, tumors such as neurofibromas can form along
nerves throughout the body, causing vision loss, hearing loss and other neurological
deficits.

Children with neurofibromatosis are at increased risk to develop central nervous
system tumors, most notably optic pathway gliomas and neurofibromas. Optic

3

pathway/ hypothalamic gliomas (OPHGs) are a relatively rare tumor type which
occurs most commonly in childhood. They represent 2%–7% of all pediatric
intracranial tumors, with 65% were observed in children younger than 5 years of age.
The majority of OPHGs are low-grade gliomas, mainly juvenile pilocytic astrocytoma
(WHO Grade I) and fibrillary astrocytoma (WHO grade 2) (Silva et al., 2000).
Although optic pathway tumors account for only 5-20% of all brain tumors in
childhood, as many as 70% of the cases are found in individuals with
neurofibromatosis Type 1 (NF1). In fact, 15% to 20% of children with
neurofibromatosis type 1 will develop an optic pathway/hypothalamic glioma
(OPHGs). In most cases the optic pathway gliomas in NF1 patients involve the
anterior visual pathway (the intraorbital optic nerve, the intracranial optic nerve, and
the optic chiasm) and they rarely extend into the optic tracts (Listernick, Charrow and
Gutmann, 1999). Gliomas of both the brain stem and the optic pathway in children
with NF1 appear to behave in a much more benign fashion than sporadic tumors in
non-NF patients. This phenomenon was first pointed out by Wright et al. in 1989
(Wright, McNab and McDonald, 1989). Listernick et al. [1994] also reported that
optic pathway gliomas in children with NF1 seldom progress once the tumors came to
medical attention (Listernick et al., 1994). Since the majority of NF1 optic pathway
gliomas behave as a benign growth, having no or very slow progression, these
patients don’t need treatment and can be monitored closely without any specific
intervention. The treatment of these asymptomatic children is only required when
there are clear signs of clinical or radiological progression. Some NF1-associated
astrocytomas grow rapidly and lead to visual loss, neurological deficits, or abnormal
endocrine function as a result of tumor extension (Arun and Gutmann, 2004) (Rutka,
Becker and Hoffman, 1997). Around one-third of patients with NF1 and optic

4

pathway tumors will become symptomatic and require treatment, usually before the
age of 5 (Silva et al., 2000).

As in most other low-grade gliomas, OPHGs can be treated with surgery, radiation
therapy (RT), chemotherapy, and the combinations of these modalities. Although
surgery serves as the primary therapy for most low-grade gliomas of childhood due to
its excellent prognosis, it has a limited role in treating chiasmal tumors, since
aggressive resection of intracranial OPHGs may cause visual, neurological and
endocrinological deficits(Sutton et al., 1995). Radiation therapy has long been the
mainstay in the treatment of chiasmal gliomas. However, its popularity was hampered
by the endocrinologic and neurocognitive side effects which are well-known
complications of radiation therapy(Erkal, Serin and Cakmak, 1997). Radiotherapy
should be particularly withheld from NF children because of significantly increased
risk of second nervous system tumors (Sharif et al., 2006) and occlusive vasculopathy
(Grill et al., 1999). Because of these concerns, chemotherapy has become increasingly
important in the treatment of these tumors due to its relatively low rates of associated
complications and low risks of longer-term late effects (Silva et al., 2000).
Chemotherapy can ultimately delay or avoid the need for radiotherapy or surgery in
young children. Currently chemotherapy is frequently the primary treatment in
patients with visual deterioration due to this disease, especially for those of young age.

Different types of chemotherapy regimens, such as lomustine, vincristine, a
combination of procarbazine (PCB), lomustine (CCNU) and vincristine (PCV), a
combination of thioguanine (TG), procarbazine (PCB), dibromodulcitol, CCNU and
vincristine (TPDCV), have been tested in LGGs to evaluate their ability to produce

5

objective response and control tumor regrowth (Gajjar et al., 1993)(Petronio et al.,
1991)(Mishra et al., 2010). The most commonly used regimens were carboplatin and
vincristine (CV), which were selected for this study for their demonstration of
effectiveness for patients with recurrent low grade gliomas. This regimen had a less
than 10% tumor progression rate within the first 12 weeks of treatment and was well
tolerated (Ater et al., 2012)(Packer et al., 1997)(Packer et al., 1993). Previous studies
enrolled very few NF1 children and evaluated treatment over relatively short median
follow-up. To evaluate the therapeutic effects and toxicity of the CV chemotherapy
regimen, a large-scale, centrally-reviewed and longer-term follow-up clinical trial
with a large number of NF1 participants would be necessary.

In this study, children with NF1 and progressive hypothalamic/optic pathway gliomas
and other progressive eligible low-grade gliomas were enrolled and treated on the
COG A9952 protocol with carboplatin and vincristine. Patients with
neurofibromatosis (NF) were non-randomly assigned to Regimen A (carboplatin and
vincristine (CV)) because they are predisposed to develop malignant myeloid
leukemia disorders and are more sensitive to alkylating agents due to genetic
background (Maris et al., 1997). The event-free survival (EFS), overall survival (OS),
and toxicity of the CV regimen was compared between NF1 patients and non-NF
patients assigned to CV. Clinical prognostic factors for EFS and OS outcome were
also identified. The results for the randomized trial for children without NF1 have
been reported separately(Ater et al., 2012).

6

CHAPTER II: METHODS

Patients
This study was performed following the Children’s Oncology Group Protocol
A9952 by member institutions of CCG (Children’s Cancer Group) in April 1997 and
POG (Pediatric Oncology Group) institutions in August 2000. New patient entry was
closed in January 2005. Patients less than 10 years of age at study entry who had
eligible low grade gliomas (WHO grades I and II) were enrolled in this study. Patients
must have had tumors with appropriate histopathology and progressive disease after
surgery, or less than 95% resection or > 1.5 cm
2
residual tumor due to incomplete
excision and in need of therapy. Eligible histopathology included low grade
astrocytoma, pilocytic astrocytoma, pleomorphic xanthoastrocytoma, subependymal
giant cell astrocytoma, infantile desmoplastic astrocytoma, low grade
oligodendroglioma, oligo-astrocytoma, ganglioglioma, and infantile desmoplastic
ganglioglioma. Chiasmatic-hypothalmic tumors intrinsic to the optic pathway were
eligible without pathological confirmation. Pathology was centrally reviewed for
eligibility by the study neuropathologist (A Yates). Tumors of all regions of the
central nervous system with appropriate histology and residual tumor were eligible,
except for intrinsic brainstem tumors of the pons and optic nerve tumors without
intracranial extension and involvement of the optic chiasm. All NF1 patients were
eligible provided that they had documented tumor progression on MRI. Patients with
clinically or radiographically progressive tumors were enrolled within 6 weeks of
progression. Chemotherapy started within 3 days of enrollment. Patients must have
received no previous treatment for their tumor other than surgery. Written informed

7

consent was obtained from all patients/guardians according to institutional and
National Cancer Institute guidelines. The protocol was approved by the institutional
review boards at all participating centers.

Study Design
Eligible patients without NF were randomly assigned to CV (carboplatin and
vincristine) or TPCV (thioguanine, procarbazine, CCNU, and vincristine). All
children with NF1 were non-randomly assigned to CV. Treatment assignment
occurred at the time when patients entered the study. Randomization was stratified by
site of disease (hypothalamic/optic vs other), status at entry (progressive LGG vs
newly diagnosed, incomplete resection) and pathology (pilocytic vs fibrillary vs
other). Only patients receiving CV treatment were included in this analysis.

Event free survival was determined by serial MRI scans and clinical examination.
MRI evaluations were performed within 6 weeks before the start of chemotherapy, at
every 3 months while on therapy and for the first year off therapy, then at 6 month
intervals for 3 years, then yearly thereafter. Baseline MRI images were reviewed for
tumor location within the optic pathway, size, presence of hydrocephalus, and
presence of tumor enhancement. The objective response to chemotherapy was
determined at 6 months and end of therapy by the institution. Two study
neuroradiologists performed central review for the baseline MRI for eligibility, the
MRI demonstrating progression and the objective response to chemotherapy. The
protocol did not include further surgery or radiation. Annual patient status regarding
events and subsequent treatment after CV continued to be reported to COG for 6.6
additional years.

8

Treatment Protocol
Chemotherapy was scheduled to start within 6 weeks of confirmatory tumor
regression for recurrent tumors, or within 6 weeks of initial tumor surgery for newly
diagnosed incompletely resected tumors considered in need of immediate therapy.
Treatment began with an induction period consisting of ten weeks of therapy followed
by two weeks of rest period. Carboplatin was administered as an IV infusion over 60
minutes at a dose of 175mg/m
2
for four consecutive weeks, followed by two weeks of
rest, then reinstituted for four more weeks. Vincristine was given as an IV bolus
infusion (maximum dose: 2.0mg) at a dose of 1.5 mg/m
2
(0.05 mg/kg if the child was
<12kg) for ten weeks, concurrent with carboplatin. After completing the induction
period, patients continued on maintenance chemotherapy when peripheral counts
recovered with the absolute neutrophil count (ANC) >1,000/ μl and platelet count
(PLT) >100,000/ μl. The maintenance period consisted of four courses, with 2 cycles
per course. Each cycle consisted of 4 weekly doses of carboplatin, three weekly doses
of vincristine (given concomitantly with the first 3 weeks of carboplatin), followed by
two weeks of rest for a total of 6 weeks. Maintenance was continued for a total of 8
cycles.

Statistical Analysis
Non-NF patients randomly assigned to carboplatin and vincristine (CV) and NF1
patients non-randomly assigned to regimen CV were compared on baseline
characteristics. Pearson chi-square test was used to test the comparability in terms of
their possible prognostic factors and stratification factors at the time of study

9

enrollment (Plackett, 1983). P values less than 0.05 were considered to indicate
statistical significance (Table 1).

Multivariate Cox regression analysis was used to analyze possible prognostic factors
for the risk of tumor recurrence (Kalbfleisch and Prentice, 2002a). Univariate Cox
regression was first used to estimate the main effect of each variable, obtaining its
hazard ratio (HR) and associated 95% confidence interval (CI). Variables with P
value less than 0.1 from univariate models were selected for inclusion in the
multivariate model. Likelihood ratio tests were used to compare the full model and the
model without single variables to obtain the global P-value for that variable. Variables
with global P-values less than 0.05 were included in the final model and considered as
significant prognostic factors (Table 2).

To analyze the cumulative probability of toxicity, the time to an event was defined as
the time from the start of therapy until the first occurrence of grade 3 or 4 toxicity, or
grade 4 only toxicity, or if none of these events occurred, until the end of
chemotherapy. Cumulative probabilities of toxicity were estimated using life table
methods: the point estimate at the end of chemotherapy was reported. A log-rank test
was used to test the equality of cumulative probabilities of toxicity between NF1 and
non-NF patients (Table 4).

The primary endpoints for analysis of treatment efficacy were event-free survival
(EFS) and overall survival (OS). The time to an event was defined as the time from
date of enrollment to first disease progression, disease recurrence, deaths from any
causes, occurrence of a second malignant neoplasm (SMN) or until last contact if no

10

events occurred. Patients with discontinued treatment due to progressive disease or
allergic reaction, or upon parent/patient request were censored at date last seen.
Overall survival was defined as the time to death from any cause. Nonparametric EFS
and overall survival curves were computed using the product-limit (Kaplan-Meier)
estimates (Kaplan and Meier, 1958), with standard errors via the Greenwood formula.
Point estimates of EFS and OS were reported as the estimate±SE. Cox regression
analysis (Cox partial likelihood ratio test) was used to test the equality of survivor
functions between NF and non-NF patients and other factors, with a P value of less
than 0.05 considered as statistically significant (Kalbfleisch and Prentice, 2002b).

All analyses were performed with the STATA Version 12 statistical package
(StataCorp. 2011. Stata Statistical Software: Release 12. College Station, TX:
StataCorp LP).

11

CHAPTER III: RESULTS

Patient Characteristics
Between April 1997 and January 2005, the COG A9952 study enrolled 428 patients.
After central pathology and chair/administrative review, 8 NF1 patients and 16 non-
NF patients were deemed to be ineligible for the study. 144 patients were excluded
from the analysis: 1 patient for lack of regimen information, 8 patients for lack of NF
status information and 135 patients for being assigned to regimen B (TPCV treatment)
(Figure 1). Characteristics of eligible patients assigned to CV at time of enrollment
are shown in Table 1. There were no significant differences between the two groups
with respect to sex (Pearson chi-square test P = 0.45) and race (Pearson chi-square
test P = 0.36), but age at diagnosis and tumor characteristics, including the amount of
residual tumor, the extent of resection, the pathology review and the tumor site,
differed significantly between NF1 and non-NF patients (Pearson chi-square test
P<0.001) (Table 1).

Treatment Failure
Of the 131 eligible NF1 patients, 43 experienced an event, as identified by their
treating institution, and 7 died. 3 NF1 patients died from progressive/persistent
disease, 1 died from infection, 1 died from hemorrhage, and 2 died from unknown
causes. Three SMNs occurred with CV-NF patients: anaplastic astrocytoma, acute
myeloid leukemia, and undifferentiated sarcoma at a median of 7.8 years (range 7.3 to
9.4 years) after diagnosis of first tumor. Two patients with SMN were first events.

12

Patients who did not experience an EFS event were followed for a median of 6.6 years
as of the cutoff date for this report (Fig 1).

Of the 129 eligible non-NF patients, 81 experienced an event, as defined by their
treating institution, and 20 died. 17 patients died from progressive/persistent disease,
and 3 died from unknown causes. One patient’s death was first event. No SMNs
occurred with CV-non-NF patients. Patients who did not experience an EFS event
were followed for a median of 6.7 years as of the cutoff date for this report (Fig 1).

Overall Outcome and Effect of NF Status
The EFS for patients with and without NF1 is shown in Figure 2, with NF1 patients
doing significantly better than non-NF patients. 43 NF1 patients had observed events
while 80 non-NF patients had observed events. Five-year EFS ± SE was 69% ± 4%
for patients with NF1 and 38% ± 5% for non-NF patients (Cox partial likelihood ratio
test P<0.001). NF1 patients also had better prognosis than non-NF patients with
regard to overall survival. Five-year OS was 98% ± 1% for patients with NF1 and 87%
± 3% for patients without NF1 (Cox partial likelihood ratio test P=0.048)
(Supplemental Fig 1).

Patients with older age did significantly better than younger patients in terms of EFS.
Of all the patients, 50 events were observed among 89 patients younger than age 3
years, 41 events among 100 patients between age 3 and 6 years, and 32 events among
71 patients older than age 6 years. Five-year EFS was 42% ± 5% for patients younger
than age 3 years, 59% ± 5% for patients between age 3 and 6 years, and 59% ± 6% for

13

patients older than age 6 years. The EFS outcomes for these 3 age groups significantly
differed (Cox partial likelihood ratio test P=0.04, log-rank trend test P=0.05) (Fig 3).

Patients with a larger amount of residual tumor (>3.0 cm
2
) had significantly worse
outcome. Twenty-seven events were observed in 72 patients with residual tumor
smaller than 1.5cm
2
, 22 events in 61 patients with residual tumor between 1.5cm
2
and
3.0 cm
2
, and 73 events for 121 patients with residual tumor bigger than 3.0cm
2
. The
five-year EFS was 65% ± 6% for patients with residual tumor smaller than 1.5cm
2
, 67%
± 6% for patients with residual tumor between 1.5cm
2
and 3.0cm
2
, and 39% ± 5% for
patients with residual tumor bigger than 3.0cm
2
. The EFS outcomes for these 3 groups
were statistically significantly different (Cox partial likelihood ratio test P=0.0004,
log-rank trend test P=0.0005) (Fig 4).

Patients with different histological type did not significantly differ in terms of their 5-
year EFS. Of all the patients, 44 events were observed in 81 patients diagnosed as
Juvenile pilocytic astrocytoma, 9 events were observed in 11 patients diagnosed as
low grade fibrillary astrocytoma, 9 events were observed in 14 patients diagnosed as
NOS low grade astrocytoma, and 3 events in 11 patients with other eligible diagnoses.
The five-year EFS was 49% ± 6% for Juvenile pilocytic astrocytoma, 33% ± 15% for
low grade fibrillary astrocytoma, 31% ± 13% for NOS low grade astrocytoma, and 73%
± 13% for other eligible diagnoses. The EFS outcomes for these 4 groups were not
statistically significantly different (Cox partial likelihood ratio test P=0.09) (Fig 5).

There was no statistical difference in 5-year EFS for patients with various extent of
tumor resection. 51 events were observed in 127 patients with no surgery, 33 events in

14

59 patients with biopsy only (<10% resection), 36 events in 64 patients with
partial/subtotal resection (10-95% resection), and 3 events in 8 patients with radical
subtotal resection (>95% resection). The five-year EFS was 60% ± 5% for patients
with no surgery, 48% ± 7% for patients with biopsy only (<10% resection), 45% ± 6%
for patients with partial/subtotal resection (10-95% resection), and 0 for patients with
radical subtotal resection (>95% resection) (For the total 8 patients in this group, 3
had events, and 5 were censored before 5 years). The EFS outcomes for these 4
groups were not statistically significantly different (Cox partial likelihood ratio test
P=0.13); however, there was a trend in cumulative survival across the ordered extent
of resection groups (log-rank trend test P=0.04) (Fig 6).

The 5-year EFS did not significantly differ by tumor site. A total of 85 events were
observed in 188 patients with optic chiasm/hypothalamic tumors, 10 events in 18
patients with thalamus tumors, 15 events in 30 patients with other supratentorial
tumors, and 13 events in 23 patients with posterior fossa/brainstem tumors. The EFS
outcomes for these 4 groups were not statistically significantly different (Cox partial
likelihood ratio test P=0.61) (Fig 7).

Since optic chiasm/hypothalamic tumors are the most prevalent subtype among NF1
patients, 5-year EFS was compared in optic chiasm/hypothalamic tumors with other
tumors. 85 events were observed in 188 patients with optic chiasm/hypothalamic
tumors, while 38 events were observed in 72 patients with other tumors. The EFS
outcomes between these two groups were not statistically significantly different (Cox
partial likelihood ratio test P=0.26) (Fig 8).

15

The EFS for optic pathway/ hypothalamic tumors with and without NF1 is shown in
Figure 9, with NF1 patients doing significantly better than non-NF patients. 38 events
were observed in 113 NF patients, while 47 events were observed in 75 non-NF
patients. Five-year EFS was 68% ± 5% for patients with NF1 and 38% ± 6% for non-
NF patients for hypothalamic/optic chiasmal tumors (Cox partial likelihood ratio test
P<0.001). NF1 patients also had better prognosis than non-NF patients with respect to
overall survival. Five-year OS was 98% ± 1% for patients with NF1 and 87% ± 4%
for hypothalamic/optic chiasmal tumors without NF1 (Cox partial likelihood ratio test
P=0.016) (Supplemental Fig 2).

The 5-year EFS differed significantly by age group among patients with optic chiasm/
hypothalamic tumors. 40 events were observed among 71 patients younger than age 3
years, 24 events among 68 patients between age 3 and 6 years, and 21 events among
49 patients older than age 6 years. Five-year EFS was 43% ± 6% for patients younger
than age 3 years, 64% ± 6% for patients between age 3 and 6 years, and 60% ± 7% for
patients older than age 6 years. The EFS outcomes for these 3 age groups were
statistically significantly different (Cox partial likelihood ratio test P=0.02, log-rank
trend test P=0.08) (Fig 10).

The amount of residual tumor was also a statistically significant predictor of 5-year
EFS outcome for patients with optic chiasm/hypothalamic tumors. 17 events were
observed in 49 patients with residual tumor smaller than 1.5cm
2
, 14 events in 46
patients with residual tumor between 1.5cm
2
and 3.0 cm
2
, and 53 events for 87
patients with residual tumor bigger than 3.0cm
2
. The five-year EFS was 68% ± 7% for
patients with residual tumor smaller than 1.5cm
2
, 71% ± 7% for patients with residual

16

tumor between 1.5cm
2
and 3.0 cm
2
, 39% ± 6% for patients with residual tumor bigger
than 3.0cm
2
. The EFS outcome for these 3 groups were statistically significantly
different (Cox partial likelihood ratio test P=0.0002, log-rank trend test P=0.0006)
(Fig 11).

Histological type was not significantly related to 5-year EFS outcome among patients
with optic chiasm/hypothalamic tumors. 25 events were observed in 50 patients
diagnosed as Juvenile pilocytic astrocytoma, 2 events were observed in 3 patients
diagnosed as low grade fibrillary astrocytoma, 3 events were observed in 5 patients
diagnosed as NOS low grade astrocytoma, and 1 event in 3 patients with other eligible
diagnoses. The five-year EFS was 54% ± 7% for Juvenile pilocytic astrocytoma, 67%
± 27% for low grade fibrillary astrocytoma, 40% ± 22% for NOS low grade
astrocytoma, and 67% ± 27% for other eligible diagnoses. The EFS outcomes for
these 4 groups were not statistically significantly different (Cox partial likelihood
ratio test P=0.87) (Fig 12).

The extent of resection was not significantly associated with 5-year EFS outcome
among patients with optic chiasm/hypothalamic tumors. 47 events were observed in
116 patients with no surgery, 20 events in 34 patients with biopsy only (<10%
resection), 18 events in 34 patients with partial/subtotal resection (10-95% resection),
and no events in 2 patients with radical subtotal resection (>95% resection). The five-
year EFS was 59% ± 5% for patients with no surgery, 45% ± 9% for patients with
biopsy only (<10% resection), 53% ± 9% for patients with partial/subtotal resection
(10-95% resection), and 100% for patients with radical subtotal resection (>95%

17

resection). The EFS outcomes for these 4 groups were not statistically significantly
different (Cox partial likelihood ratio test P=0.20, log-rank trend test P=0.20) (Fig 13).

We also compared the 5-year EFS rates for different amounts of residual tumor,
stratified by NF status. For NF1 patients, five-year EFS was 72% ± 7% for patients
with residual tumor smaller than 1.5cm
2
, 83% ± 6% for patients with residual tumor
between 1.5cm
2
and 3.0 cm
2
, 51% ± 8% for patients with residual tumor bigger than
3.0cm
2
. The EFS outcome for these 3 groups were statistically significantly different
(Cox partial likelihood ratio test P=0.0058, log-rank trend test P=0.03). The survivor
functions for the 3 groups were not statistically different among non-NF patients (Cox
partial likelihood ratio test P=0.64) (Fig 14). The amount of residual tumor was a
significant predictor of 5-year EFS outcome for NF1 patients, but not for non-NF
patients.

In contrast, age was a significant predictor of 5-year EFS outcome for non-NF
patients, but not for NF1 patients. For non-NF patients, five-year EFS was 12% ± 5%
for patients younger than age 3 years, 51% ± 7% for patients between age 3 and 6
years, 44% ± 9% for patients older than age 6 years. The EFS outcomes for these 3
age groups were statistically significantly different (Cox partial likelihood ratio test
P=0.0008, log rank trend test P=0.0027). The survivor functions for the 3 age groups
were not statistically different for NF1 patients (Cox partial likelihood ratio test
P=0.91) (Fig 15).

Similarly, tumor histology also was significantly associated with 5-year EFS outcome
for non-NF patients, but not for NF1 patients. For non-NF patients, the five-year EFS

18

was 45% ± 6% for Juvenile pilocytic astrocytoma, 15% ± 13% for low grade fibrillary
astrocytoma, 20% ± 13% for NOS low grade astrocytoma, and 73% ± 13% for other
eligible diagnoses. The EFS outcomes for these 4 groups were statistically
significantly different (Cox partial likelihood ratio test P=0.047). The survivor
functions for the 4 groups were not statistically different for NF1 patients (Cox partial
likelihood ratio test P=0.99) (Fig 16).

Comparing optic chiasm/hypothalamic tumors with others, the five-year EFS was 68%
± 5% for NF1 patients with optic chiasm/hypothalamic tumors versus 77% ± 10% for
NF1 patients with other tumors. The five-year EFS was 38% ± 6% for non-NF
patients with optic chiasm/hypothalamic tumors versus 39% ± 7% for non-NF
patients with other tumors. The survivor functions between these two groups were not
statistically significantly different for both NF1 and non-NF patients (Cox partial
likelihood ratio test P=0.48 and P=0.98, respectively) (Fig 17).

Prognostic Factors
Multivariate analysis revealed four factors that were independently predictive of EFS:
NF status, age, the amount of residual tumor and the histological type by pathology
review, as shown in Table 2. Non-NF patients were 4.33 times more likely (95% CI,
1.92 to 9.81) to have progression/relapse events compared with NF1 patients. The
relative risk for progression/relapse was 0.54 times lower in patients who were
between age 3 to 6 years (95% CI, 0.35 to 0.84) than in those who were younger than
age 3 years; the risk was 0.60 times lower in patients who were older than 6 years (95%
CI, 0.38 to 0.96) than those who were younger than age 3 years. The relative risk for
progression/relapse was 2.39 times higher (95% CI, 1.21 to 4.74) in patients with

19

residual tumor bigger than 3 cm
2
than in patients with residual tumor less than 1.5 cm
2
.
Patients who were diagnosed with low grade fibrillary astrocytoma had 2.49 times
higher risk (95% CI, 1.18 to 5.26) for progression/relapse than patients who were
diagnosed with Juvenile pilocytic astrocytoma. The interaction term between NF
status and residual tumor was significantly associated with the EFS outcome (P=0.029)
(Table 3). Non-NF patients with residual tumor between 1.5cm
2
and 3.0 cm2 were
4.57 times more likely to experience progression/relapse compared to NF1 patients
with residual tumor less than 1.5 cm
2
. Non-NF patients with residual tumor bigger
than 3.0 cm
2
were 5.23 times more likely to experience progression/relapse compared
to NF1 patients with residual tumor less than 1.5 cm
2
. In NF1 patients, children with
residual tumor bigger than 3.0 cm
2
were 2.05 times more likely to experience
progression/relapse compared to those with residual tumor less than 1.5 cm
2
(95% CI,
1.04 to 4.04); while in non-NF patients, children with residual tumor bigger than 3.0
cm
2
were 1.28 times more likely to experience progression/relapse compared to those
with residual tumor less than 1.5 cm
2
(95% CI, 0.70 to 2.35). Other variables
including sex, race, extent of resection and tumor site were not statistically associated
with 5-year EFS outcome (P>0.05). The amount of residual tumor was the only
statistically significant prognostic factor of OS (Supplemental Table 1).

Toxicity
The cumulative probability of toxicity by end of chemotherapy is shown in Table 4.
All allergic reactions reported were attributed to carboplatin. When the study initially
opened, patients were required to be removed from carboplatin therapy for any grade
of allergic reaction. On August 11, 2000, the protocol was amended to allow patients
with grade 1-2 allergic reactions to remain on study as long as they did not progress to

20

grade 3-4. A total of 16 NF1 and 24 non-NF patients went off therapy for toxicity
reactions. The most commonly reported toxic effects were absolute neutrophil count
(ANC), platelets (PLT), hemoglobin (HGB), peripheral nervous system, central
nervous system, allergy and infection. Compared to non-NF patients, NF1 patients
had statistically significant lower probability of experiencing ANC toxicity, regardless
of grade 3 or 4 (log rank test P value=0.03) or grade 4 only (log rank test P
value=0.01). NF1 patients also had significantly lower probability of experiencing
grade 3 or 4 HGB toxicity (log rank test P value=0.0058). Other toxic effects didn’t
significantly differ between NF1 and non-NF patients (log rank test P value>0.05)
(Table 4).

Fig1

1: CONSOR RT diagram m

21

22

Fig2. Event-free survival (EFS) for NF1 and non-NF children. 131 NF1 patients and
129 non-NF patients were included in the analysis. 43 NF1 patients had observed
events while 80 non-NF patients had observed events. NF1 patients had significantly
better prognosis than non-NF patients in terms of EFS (Cox partial likelihood ratio
test P<0.001)

0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
NF
non-NF
Kaplan-Meier survival estimates

23

Fig 3. Event-free survival by age at study entry. 50 events were observed among 89
patients younger than age 3 years, 41 events among 100 patients between age 3 and 6
years and 32 events among 71 patients older than age 6 years. The EFS outcomes for
these 3 age groups were statistically significantly different (Cox partial likelihood
ratio test P=0.04, log-rank trend test P=0.05).

0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
<3 years
3-6 years
>=6 years
EFS by Age

24

Fig 4. Event-free survival by the amount of residual tumor at study entry. 27 events
were observed in 72 patients with residual tumor smaller than 1.5cm
2
, 22 events in 61
patients with residual tumor between 1.5cm
2
and 3.0 cm
2
, and 73 events in 121
patients with residual tumor bigger than 3.0cm
2
. The EFS outcomes for these 3 groups
were statistically significantly different (Cox partial likelihood ratio test P=0.0004,
log-rank trend test P=0.0005).

0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
<1.5 cm2
1.5-3.0 cm2
>3.0 cm2
EFS by Residual Tumor

25

Fig 5. Event-free survival by pathology review at study entry. 44 events were
observed in 81 patients diagnosed as Juvenile pilocytic astrocytoma, 9 events were
observed in 11 patients diagnosed as low grade fibrillary astrocytoma, 9 events were
observed in 14 patients diagnosed as NOS low grade astrocytoma, and 3 events in 11
patients with other eligible diagnoses. The EFS outcomes for these 4 groups were not
statistically significantly different (Cox partial likelihood ratio test P=0.09).

0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
Juvenile pilocytic astro
Low grade fibrillary astro
Low grade astrocytoma, NOS
Other eligible diagnosis
EFS by pathology review

26

Fig 6. Event-free survival by extent of resection. 51 events were observed in 127
patients with no surgery, 33 events in 59 patients with biopsy only (<10% resection),
36 events in 64 patients with partial/subtotal resection (10-95% resection), and 3
events in 8 patients with radical subtotal resection (>95% resection). The EFS
outcomes for these 4 groups were not statistically significantly different (Cox partial
likelihood ratio test P=0.13). There was a trend in cumulative survival across the
ordered extent of resection groups (log-rank trend test P=0.04)

0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
No surgery
Biopsy only (<10%)
Partial/Subtotal (10-95%)
Radical Subtotal (>95%)
EFS by extent of resection

27

Fig 7. Event-free survival by tumor site. 85 events were observed in 188 patients with
optic chiasm/hypothalamic tumors, 10 events in 18 patients with thalamus tumors, 15
events in 30 patients with other supratentorial tumors, 13 events in 23 patients with
posterior fossa/brainstem tumors. The survivor functions for these 4 groups were not
statistically significantly different (Cox partial likelihood ratio test P=0.61)

0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
Optic chiasm/hypothalamic
Thalamus
Other supratentorial
Posterior fossa/brainstem
EFS by tumorsite

28

Fig 8. Event-free survival by tumor site. 85 events were observed in 188 patients with
optic chiasm/hypothalamic tumors, while 38 events were observed in 72 patients with
other tumors. The survivor functions between these two groups were not statistically
significantly different (Cox partial likelihood ratio test P=0.26).

0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
Optic chiasm/hypothalamic
Others
EFS by tumorsite

29

Fig 9. Event-free survival for NF1 or non-NF children with optic pathway/
hypothalamic tumors. 38 events were observed in 113 NF1 patients while 47 events
were observed in 75 non-NF patients. NF1 patients had significantly better prognosis
than non-NF patients for optic pathway/hypothalamic tumors in terms of EFS (Cox
partial likelihood ratio test P<0.001).

0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
NF
non-NF
Kaplan-Meier survival estimates

30

Fig 10. Event-free survival by age at study entry for patients with optic pathway/
hypothalamic tumors. 40 events were observed among 71 patients younger than age 3
years, 24 events among 68 patients between age 3 and 6 years, and 21 events among
49 patients older than age 6 years. The EFS outcomes for these 3 age groups were
statistically significantly different (Cox partial likelihood ratio test P=0.02, log-rank
trend test P=0.08).

0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
<3 years
3-6 years
>=6 years
EFS by Age

31

Fig 11. Event-free survival by amount of residual tumor at study entry for patients
with optic pathway/ hypothalamic tumors. 17 events were observed in 49 patients
with residual tumor smaller than 1.5cm
2
, 14 events in 46 patients with residual tumor
between 1.5cm
2
and 3.0 cm
2
, and 53 events for 87 patients with residual tumor bigger
than 3.0cm
2
. The EFS outcomes for these 3 groups were statistically significantly
different (Cox partial likelihood ratio test P=0.0002, log-rank trend test P=0.0006).

0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
<1.5 cm2
1.5-3.0 cm2
>3.0 cm2
EFS by Residual Tumor

32

Fig 12. Event-free survival by pathology review for patients with optic pathway/
hypothalamic tumors. 25 events were observed in 50 patients diagnosed as Juvenile
pilocytic astrocytoma, 2 events were observed in 3 patients diagnosed as low grade
fibrillary astrocytoma, 3 events were observed in 5 patients diagnosed as NOS low
grade astrocytoma, and 1 event in 3 patients with other eligible diagnoses. The
survivor functions for these 4 groups were not statistically significantly different (Cox
partial likelihood ratio test P=0.87).

0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
Juvenile pilocytic astro
Low grade fibrillary astro
Low grade astrocytoma, NOS
Other eligible diagnosis
EFS by pathology review

33

Fig 13. Event-free survival by extent of resection for patients with optic pathway/
hypothalamic tumors. 47 events were observed in 116 patients with no surgery, 20
events in 34 patients with biopsy only (<10% resection), 18 events in 34 patients with
partial/subtotal resection (10-95% resection), and no events in 2 patients with radical
subtotal resection (>95% resection). The survivor functions for these 4 groups were
not statistically significantly different (Cox partial likelihood ratio test P=0.20, log-
rank trend test P=0.20).

0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
No surgery
Biopsy only (<10%)
Partial/Subtotal (10-95%)
Radical Subtotal (>95%)
EFS by extent of resection

34

Fig14. The amount of residual tumor as a prognostic factor in NF1 or non-NF patients.
The EFS outcomes for these 3 groups were statistically significantly different for NF1
patients (Cox partial likelihood ratio test P=0.0058, log-rank trend test P=0.03), but
were not significantly different for non-NF patients (Cox partial likelihood ratio test
P=0.64).
0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
<1.5 cm2
1.5-3.0 cm2
>3.0 cm2
EFS by Residual Tumor for NF1
0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
<1.5 cm2
1.5-3.0 cm2
>3.0 cm2
EFS by Residual Tumor for non-NF

35

Fig15. Age as a prognostic factor in NF1 or non-NF patients. The survivor functions
for the 3 age groups were not statistically significantly different for NF1 patients (Cox
partial likelihood ratio test P=0.91), but were statistically significantly different for
non-NF patients (Cox partial likelihood ratio test P=0.0008, log rank trend test
P=0.0027).

0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
<3 years
3-6 years
>=6 years
EFS by Age for NF1
0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
<3 years
3-6 years
>=6 years
EFS by Age for non-NF Children

36

Fig16. Pathology review as a prognostic factor in NF1 or non-NF patients. The
survivor functions for the 4 groups were not statistically different for NF1 patients
(Cox partial likelihood ratio test P=0.99), but were statistically significantly different
for non-NF patients (Cox partial likelihood ratio test P=0.047).

0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
Juvenile pilocytic astro
Low grade fibrillary astro
Low grade astrocytoma, NOS
EFS by pathology review for NF1
0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
Juvenile pilocytic astro
Low grade fibrillary astro
Low grade astrocytoma, NOS
Other eligible diagnosis
EFS by pathology review for non-NF

37

Fig17. Tumor site as a prognostic factor in NF1 or non-NF patients. The survivor
functions between these two groups were not statistically significantly different for
both NF1 and non-NF patients (Cox partial likelihood ratio test P=0.48 and P=0.98,
respectively).

0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
optic chiasm/hypothalamic
Others
EFS by Tumor Site for NF1
0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
optic chiasm/hypothalamic
Others
EFS by Tumor Site for non-NF Children

38

Table 1: Patient Characteristics

CV-NF (n=131) CV (n=129) p-value
Characteristic No. % No. %
Sex
Male
Female

66
65

50%
50%

59
70

46%
54%

0.45
Age (years)
<1
1 to 5
5 to 10

3
76
52

2%
58%
40%

19
61
49

15%
47%
38%

0.001
Race
White
Hispanic
African American
Other/unknown

110
9
8
4

84%
7%
6%
3%

99
13
8
9

77%
10%
6%
7%

0.36
Amount of residual tumor
≤1.5 cm2
1.5 - 3.0 cm2
>3 cm2
Unknown

47
36
43
5

36%
27%
33%
4%

25
25
78
1

19.4%
19.4%
60.5%
0.8%

<0.001
Extent of resection
No surgery
Biopsy only (<10%)
Partial/Subtotal (10-95%)
Radical Subtotal (>95%)
Unknown

106
13
8
2
2

81%
10%
6%
1.5%
1.5%

21
46
56
6
0

16%
36%
43%
5%
0%

<0.001
Pathology review
Juvenile pilocytic astro
Low grade fibrillary astro
Low grade astrocytoma, NOS
Other eligible diagnosis
No surg./missing/insufficient

14
2
3
0
112

11%
1.5%
2%
0%
85.5%

67
9
11
11
31

52%
7%
8.5%
8.5%
24%

<0.001
Tumor site
Optic chiasm/hypothalamic
Thalamus
Other supratentorial
Posterior fossa/brainstem
Unknown/missing

113
3
11
3
1

86%
2%
9%
2%
1%

75
15
19
20
0

58%
12%
15%
15%
0%

<0.001
Tumor site2
Optic chiasm/hypothalamic
Others

113
18

86%
14%

75
54

58%
42%

<0.001

39

Table 2: Multivariate risk factors affecting EFS outcome
Univariate Models Multivariate models
Characteristic Hazard
Ratio
p-value 95%CI Hazard
Ratio
p-value 95%CI
Sex
Male
Female

1
1.09
0.64

----
0.76-1.55
Not
included

----

----

Age (years)
<3
3 to 6
>=6

1
0.61
0.66
0.04

----
0.40-0.92
0.42-1.03

1
0.54
0.60
0.03

----
0.35-0.84
0.38-0.96
Race
White
Hispanic
African American
Other/unknown

1
1.13
0.61
0.92
0.66

----
0.61-2.11
0.25-1.51
0.40-2.10
Not
included

----

----

Amount of residual tumor
≤1.5 cm2
1.5 - 3.0 cm2
>3 cm2
Unknown

1
0.92
1.98
0.36
0.0003

----
0.52-1.61
1.28-3.09
0.05-2.67

1
0.62
2.39
0.55
0.017

----
0.25-1.53
1.21-4.74
0.07-4.19
Extent of resection
No surgery
Biopsy only (<10%)
Partial/Subtotal (10-95%)
Radical Subtotal (>95%)
Unknown

1
1.50
1.59
1.26
-----
0.076

----
0.97-2.32
1.04-2.44
0.39-4.06
-----
Not
included

----

Pathology review
Juvenile pilocytic astro
Low grade fibrillary astro
Low grade astrocytoma, NOS
Other eligible diagnosis
No surg./missing/insufficient

1
1.89
1.55
0.50
0.71
0.02

----
0.92-3.88
0.76-3.19
0.15-1.60
0.48-1.05

1
2.49
1.71
0.40
1.75
0.008

----
1.18-5.26
0.81--3.58
0.12-1.29
1.08-2.84
Tumor site
Optic chiasm/hypothalamic
Thalamus
Other supratentorial
Posterior fossa/brainstem
Unknown/missing

1
1.43
1.15
1.33
-----
0.53

----
0.74-2.75
0.67-2.00
0.74-2.39
-----
Not
included

----

---

Tumor site 2
Optic chiasm/hypothalamic
Others

1
1.25
0.26
-----
0.85-1.83
Not
included

-----

-----
NF status
NF
Non-NF

1
2.60
<0.001
----
1.79-3.77

1
4.33
<0.001
----
1.92-9.81
NF status х residual tumor
NF х residual tumor ( ≤1.5 cm
2
)
Non-NF хresidual tumor(1.5 - 3.0 cm
2
)
Non-NF х residual tumor (>3 cm
2
)
Non-NF х residual tumor (Unknown)

1
1.77
0.63
----
0.26

-----
0.55-5.72
0.26-1.58
----

1
1.71
0.50
----
0.029

----
0.52-5.63
0.20-1.28
----
NF status хiage2
NF х age (<3)
Non-NF х age (3 to 6)
Non-NF х age (>=6)

1
0.38
0.40
0.057

-----
0.15-0.92
0.16-1.01
Not
included

40

Table 3. Interaction between NF status and amount of residual tumor

NF status х pathology review
NF х Juvenile pilocytic astro
Non-NF х Low grade fibrillary astro
Non-NF хLow grade astrocytoma, NOS
Non-NFх Other eligible diagnosis
Non-NFх No surg./missing/insufficient

1
1.04
2.18
1
2.83
0.22

-----
0.17-6.52
0.22-21.34
----
0.96-8.27
Not
included

NF Non‐NF
Hazard Ratio Hazard Ratio
Amount of residual tumor
≤1.5 cm
2

1.5 ‐ 3.0 cm
2

>3 cm
2

1
0.62
2.40
4.33
4.57
5.23

41

Table 4: Cumulative probability of toxicity by End of Chemotherapy

Key Toxicity
Grade 3 or 4
CV CV-NF p-value*
Grade 4 only
CV CV-NF p-value*
Absolute neutrophil count 0.94 0.92 0.03 0.72 0.62 0.01
Platelets 0.21 0.17 0.42 0.08 0.05 0.49
Hemoglobin 0.34 0.19 0.01 0.06 0.05 0.45
Alanine transaminase 0.02 0.03 0.44 0.01 0.01 0.94
Total bilirubin 0 0.01 0.33 0 0.01 0.33
Creatinine 0.01 0 0.28 0.01 0 0.28
Creatinine clearance 0 0 ---- 0 0 ----
Pulmonary 0.02 0.02 0.93 0.01 0.01 0.94
Calcium 0.05 0.03 0.64 0.03 0.02 0.93
Magnesium 0.03 0.02 0.95 0.01 0.02 0.57
Peripheral nervous system 0.19 0.22 0.54 0 0 -----
Central nervous system 0.12 0.05 0.08 0.01 0 0.31
Allergy 0.10 0.08 0.40 0.02 0.06 0.13
Infection 0.22 0.19 0.58 0 0.01 0.32

*Log-rank test was used to compare the cumulative probabilities of toxicity between
NF1 and non-NF patients

42

Supplemental Fig 1. Overall survival (OS) in NF1 or non-NF patients. NF1 patients
did significantly better than non-NF patients in terms of OS (Cox partial likelihood
ratio test P=0.048).

0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
NF
non-NF
Kaplan-Meier survival estimates

43

Supplemental Fig 2. The Overall survival (OS) for optic pathway/ hypothalamic
tumors in NF1 or non-NF patients. NF1 patients with optic pathway/ hypothalamic
tumors did significantly better than non-NF patients in terms of OS (Cox partial
likelihood ratio test P=0.016).

0.00 0.25 0.50 0.75 1.00
Probability
0 2 4 6 8 10 12 14
Years from study entry
NF
non-NF
Kaplan-Meier survival estimates

44

Supplemental Table 1: Multivariate risk factors affecting OS outcome

Univariate Models Multivariate models
Characteristic Hazard
Ratio
p-value 95%CI Hazard
Ratio
p-value 95%CI
Sex
Male
Female

1
1.95
0.09

----
0.88-4.34
Not
included

----

----

Age (years)
<3
3 to 6
>=6

1
0.35
0.47
0.056

----
0.14-0.91
0.18-1.22
Not
included

Race
White
Hispanic
African American
Other/unknown

1
0.83
1.33
-----
0.37

----
0.20-3.51
0.31-5.68
-----
Not
included

----

----

Amount of residual tumor
≤1.5 cm2
1.5 - 3.0 cm2
>3 cm2
Unknown

1
0.16
1.94
-----
0.0007

----
0.02-1.35
0.78-4.85
-----

1
0.14
1.41
----
0.01

----
0.02-1.19
0.54-3.67
-----
Extent of resection
No surgery
Biopsy only (<10%)
Partial/Subtotal (10-95%)
Radical Subtotal (>95%)
Unknown

1
2.64
1.83
-----
-----
0.19

----
1.07-6.50
0.70-4.76
-----
-----
Not
included

----

Pathology review
Juvenile pilocytic astro
Low grade fibrillary astro
Low grade astrocytoma, NOS
Other eligible diagnosis
No surg./missing/insufficient

1
2.54
1.38
1.86
0.65
0.31

----
0.69-9.41
0.30-6.42
0.40-8.64
0.27-1.58
Not
included

Tumor site
Optic chiasm/hypothalamic
Thalamus
Other supratentorial
Posterior fossa/brainstem
Unknown/missing

1
2.74
1.43
0.48
-----
0.41

----
0.92-8.13
0.48-4.22
0.07-3.67
-----
Not
included

----

---

Tumor site2
Optic chiasm/hypothalamic
Others

1
1.40
0.42
-----
0.63-3.13

NF status
NF
Non-NF

1
3.17
0.005
----
1.34-7.51
Not
included

45

CHAPTER IV: DISCUSSION

This study was part of A9952 clinical trial which was started in 1997 to validate the
approach of chemotherapy for low-grade gliomas to improve survival and delay the
needs for radiotherapy for young children. A large number of NF1 children were
enrolled in the study and received a single chemotherapy regimen in this prospective,
centrally reviewed clinical trial. NF1 patients were not randomized because of
concerns for an increased risk of second malignancies due to the genetic background
(Maris et al., 1997). This study evaluated the safety, tolerability and efficacy of
carboplatin and vincristine in pediatric patients and validated this regimen as a
promising chemotherapy for low-grade gliomas in NF1 patients younger than 10
years of age.

As shown in Table 1, some characteristics of NF1 patients and non-NF patients also
receiving CV differed. For example, the NF1 sample included fewer children less than
1 year old (2% versus 15% in non-NF patients). They also had lower percentage of
residual tumors larger than 3cm
2
(33% versus 60.5% in non-NF patients). These
differences may partly explain why NF patients had better prognosis than non-NF.
However, after adjustment for age and the amount of residual tumor, the superiority
of NF1 over non-NF patients in terms of EFS outcome still existed, which suggested
that NF1 patients did better than non-NF patients due to their intrinsic genetic
background.

46

The NF1 patients demonstrated good tolerability for the CV chemotherapy. They had
statistically significantly lower risk to experience absolute neutrophil count (ANC)
and hemoglobin (HGB) toxicity compared to non-NF children. There were total 3
second malignancies observed in this study, all of which occurred among NF1
patients. It is worth noting that these three patients with SMNs all received subsequent
treatment with temozolomide, an alkylator with potential risk for second neoplasms. It
is consistent with the observation from Duke University investigators, which reported
that 2 NF1 patients were transformed from juvenile pilocytic astrocytoma (JPA) to
anaplastic pilocytic astrocytoma (APA) after chemotherapy treatment with carboplatin,
vincristine and oral temozolomide(Peters, Cummings and Gururangan, 2011). This
result suggests that temozolomide should be used with caution in NF1 patients.

This study suggested that the prognostic factors were similar among patients with all
types of tumors and patients with optic chiasm/hypothalamic tumors only. The factors
associated with better EFS prognosis included age greater than 3 years (relative less
than 3 years), NF1 status (relative to non-NF), amount of residual tumor smaller than
3cm
2
(relative to larger residual tumor). This result is consistent with some previous
studies(Silva et al., 2000)(Ater et al., 2012). Interestingly, it is contrary to an earlier
study of CV therapy on NF1 and non-NF patients, which claimed that there was no
statistical difference in progression-free survival rates between NF1 and non-NF
children, and children younger than 5 years had better prognosis than older
children(Packer et al., 1997). This difference may be partly explained by their
different study population, smaller sample size, modified study design and therapy
treatment schedule. We also noticed that the amount of residual tumor was a
significant predictor for NF1 patients but not for non-NF patients, while age and

47

histological subtype were significant predictors for non-NF patients but not for NF
patients. These findings could serve as a basis for possible stratification in future
clinical trials and help to develop more targeted therapy for individuals.

Neurofibromatosis type 1 (NF1) is a common inherited tumor syndrome in which
affected individuals are prone to develop benign and malignant tumors.
Understanding the molecular mechanism of NF1 disease may help to improve
regimens for chemotherapy. Several pathways are involved in the development of
tumors associated with NF1, among which the most clearly elucidated is that
neurofibromin functions as a tumor suppressor by modulating the p21-ras signaling
pathway(Listernick et al., 1999). Mutations of the NF1 suppressor gene lead to
functional loss of neurofibromin protein which increases Ras activity and results in
increased cell proliferation and tumor formation (Arun and Gutmann, 2004)(Brems et
al., 2009). The current management of such malignancies is unsatisfactory. It is of
great significance to further explore the molecular changes underlying tumor
formation and progression to identify potential therapeutic targets (Dilworth et al.,
2006). Specific targeted therapies may allow in the future for more individualized
therapeutic regimens. A broad spectrum of molecular biological techniques and
bioinformatics tools can work together to analyze every tumor for each patient and
identify and validate predictive and pharmacodynamic (PD) biomarkers to stratify
patients for clinical trials. These findings will bridge the gap between treatment given
to patients today and the future goal of personalized medicine.

In conclusion, this study validated carboplatin and vincristine as an efficacious
chemotherapy for pediatric patients younger than 10 years of age. Patients with NF1

48

genetic background, older age and smaller amount of residual tumor had statistically
significantly better prognosis. NF1 patients did significantly better than non-NF
patients in terms of 5-year event-free survival and overall survival. They also had
better tolerability for this regimen than non-NF patients. Further improvement of
chemotherapy regimens for NF1 patients is still needed for targeted therapy and more
personalized medicine.

49

REFERENCES
Arun, D. and Gutmann, D. (2004) Recent advances in neurofibromatosis type 1.
Current Opinion in Neurology April 2004, 17, 101–105.
Ater, J.L., Zhou, T., Holmes, E., Mazewski, C.M., Booth, T.N., Freyer, D.R., et al.
(2012) Randomized Study of Two Chemotherapy Regimens for Treatment of
Low-Grade Glioma in Young Children: A Report From the Children’s
Oncology Group. Journal of Clinical Oncology, 30, 2641–2647.
Brems, H., Beert, E., de Ravel, T. and Legius, E. (2009) Mechanisms in the
pathogenesis of malignant tumours in neurofibromatosis type 1. The Lancet
Oncology, 10, 508–515.
Campagna, M., Opocher, E., Viscardi, E., Calderone, M., Severino, S.M., Cermakova,
I., et al. (2010) Optic pathway glioma: Long-term visual outcome in children
without neurofibromatosis type-1. Pediatric Blood & Cancer, 55, 1083–1088.
Dilworth, J.T., Kraniak, J.M., Wojtkowiak, J.W., Gibbs, R.A., Borch, R.F., Tainsky,
M.A., et al. (2006) Molecular targets for emerging anti-tumor therapies for
neurofibromatosis type 1. Biochemical Pharmacology, 72, 1485–1492.
Erkal, H.S., Serin, M. and Cakmak, A. (1997) Management of optic pathway and
chiasmatic-hypothalamic gliomas in children with radiation therapy.
Radiotherapy and Oncology: Journal of the European Society for Therapeutic
Radiology and Oncology, 45, 11–15.
Farwell, J.R., Dohrmann, G.J. and Flannery, J.T. (1977) Central nervous system
tumors in children. Cancer, 40, 3123–3132.
Fisher, M.J., Loguidice, M., Gutmann, D.H., Listernick, R., Ferner, R.E., Ullrich, N.J.,
et al. (2012) Visual outcomes in children with neurofibromatosis type 1-
associated optic pathway glioma following chemotherapy: a multicenter
retrospective analysis. Neuro-Oncology, 14, 790–797.
Gajjar, A., Heideman, R.L., Kovnar, E.H., Langston, J.A., Sanford, R.A., Douglass,
E.C., et al. (1993) Response of pediatric low grade gliomas to chemotherapy.
Pediatric Neurosurgery, 19, 113–118; discussion 119–120.
Gajjar, A., Sanford, R.A., Heideman, R., Jenkins, J.J., Walter, A., Li, Y., et al. (1997)
Low-grade astrocytoma: a decade of experience at St. Jude Children’s
Research Hospital. Journal of Clinical Oncology: Official Journal of the
American Society of Clinical Oncology, 15, 2792–2799.
Goodden, J., Pizer, B., Pettorini, B., Williams, D., Blair, J., Didi, M., et al. (2014) The
role of surgery in optic pathway/hypothalamic gliomas in children. Journal of
Neurosurgery. Pediatrics, 13, 1–12.
Grill, J., Couanet, D., Cappelli, C., Habrand, J.L., Rodriguez, D., Sainte-Rose, C., et
al. (1999) Radiation-induced cerebral vasculopathy in children with

50

neurofibromatosis and optic pathway glioma. Annals of Neurology, 45, 393–
396.
Kalbfleisch, J.D. and Prentice, R.L. (2002a) Relative Risk (Cox) Regression Models.
The Statistical Analysis of Failure Time Data pp. 95–147. John Wiley & Sons,
Inc.
Kalbfleisch, J.D. and Prentice, R.L. (2002b) Failure Time Models. The Statistical
Analysis of Failure Time Data pp. 31–51. John Wiley & Sons, Inc.
Kalin-Hajdu, E., Décarie, J.-C., Marzouki, M., Carret, A.-S. and Ospina, L.H. (2014)
Visual acuity of children treated with chemotherapy for optic pathway gliomas.
Pediatric Blood & Cancer, 61, 223–227.
Kaplan, E.L. and Meier, P. (1958) Nonparametric Estimation from Incomplete
Observations. Journal of the American Statistical Association, 53, 457.
Listernick, R., Charrow, J. and Gutmann, D.H. (1999) Intracranial gliomas in
neurofibromatosis type 1. American Journal of Medical Genetics, 89, 38–44.
Maris, J.M., Wiersma, S.R., Mahgoub, N., Thompson, P., Geyer, R.J., Hurwitz, C.G.,
et al. (1997) Monosomy 7 myelodysplastic syndrome and other second
malignant neoplasms in children with neurofibromatosis type 1. Cancer, 79,
1438–1446.
Meister, B. (2011) Conservative treatment of paediatric hypothalamic and optic
pathway gliomas. memo - Magazine of European Medical Oncology, 4, 14–15.
Merchant, T.E., Conklin, H.M., Wu, S., Lustig, R.H. and Xiong, X. (2009) Late
effects of conformal radiation therapy for pediatric patients with low-grade
glioma: prospective evaluation of cognitive, endocrine, and hearing deficits.
Journal of Clinical Oncology: Official Journal of the American Society of
Clinical Oncology, 27, 3691–3697.
Mishra, K.K., Squire, S., Lamborn, K., Banerjee, A., Gupta, N., Wara, W.M., et al.
(2010) Phase II TPDCV protocol for pediatric low-grade
hypothalamic/chiasmatic gliomas: 15-year update. Journal of Neuro-Oncology,
100, 121–127.
Packer, R.J., Ater, J., Allen, J., Phillips, P., Geyer, R., Nicholson, H.S., et al. (1997)
Carboplatin and vincristine chemotherapy for children with newly diagnosed
progressive low-grade gliomas. Journal of Neurosurgery, 86, 747–754.
Packer, R.J., Lange, B., Ater, J., Nicholson, H.S., Allen, J., Walker, R., et al. (1993)
Carboplatin and vincristine for recurrent and newly diagnosed low-grade
gliomas of childhood. Journal of Clinical Oncology: Official Journal of the
American Society of Clinical Oncology, 11, 850–856.
Peters, K.B., Cummings, T.J. and Gururangan, S. (2011) Transformation of juvenile
pilocytic astrocytoma to anaplastic pilocytic astrocytoma in patients with
neurofibromatosis type I. Journal of Pediatric Hematology/Oncology, 33,
e198–201.

51

Petronio, J., Edwards, M.S., Prados, M., Freyberger, S., Rabbitt, J., Silver, P., et al.
(1991) Management of chiasmal and hypothalamic gliomas of infancy and
childhood with chemotherapy. Journal of Neurosurgery, 74, 701–708.
Plackett, R.L. (1983) Karl Pearson and the Chi-Squared Test. International Statistical
Review / Revue Internationale de Statistique, 51, 59.
Rutka, J.T., Becker, L.E. and Hoffman, H.J. (1997) Optic Chiasmatic-Hypothalamic
Glioma. Brain Pathology, 7, 799–806.
Sharif, S., Ferner, R., Birch, J.M., Gillespie, J.E., Gattamaneni, H.R., Baser, M.E., et
al. (2006) Second primary tumors in neurofibromatosis 1 patients treated for
optic glioma: substantial risks after radiotherapy. Journal of Clinical Oncology:
Official Journal of the American Society of Clinical Oncology, 24, 2570–2575.
Shofty, B., Ben-Sira, L., Freedman, S., Yalon, M., Dvir, R., Weintraub, M., et al.
(2011) Visual outcome following chemotherapy for progressive optic pathway
gliomas. Pediatric Blood & Cancer, 57, 481–485.
Sievert, A.J. and Fisher, M.J. (2009) Pediatric Low-Grade Gliomas. Journal of Child
Neurology, 24, 1397–1408.
Silva, M.M., Goldman, S., Keating, G., Marymont, M.A., Kalapurakal, J. and Tomita,
T. (2000) Optic Pathway Hypothalamic Gliomas in Children under Three
Years of Age: The Role of Chemotherapy. Pediatric Neurosurgery, 33, 151–8.
Sutton, L.N., Molloy, P.T., Sernyak, H., Goldwein, J., Phillips, P.L., Rorke, L.B., et al.
(1995) Long-term outcome of hypothalamic/chiasmatic astrocytomas in
children treated with conservative surgery. Journal of Neurosurgery, 83, 583–
589.
Wright, J.E., McNab, A.A. and McDonald, W.I. (1989) Optic nerve glioma and the
management of optic nerve tumours in the young. The British Journal of
Ophthalmology, 73, 967–974.

Abstract (if available)

Abstract To evaluate and compare the therapeutic effects and toxicity of chemotherapy for low grade gliomas (LGG) in Neurofibromatosis Type 1 (NF1) and non-NF children, patients less than 10 years of age with eligible progressive LGG or with incomplete excision of primary LGG requiring immediate treatment were enrolled and treated on the COG A9952 protocol with carboplatin and vincristine (CV). NF1 children tolerated CV well and had a superior event-free survival and overall survival compared to CV-treated children without NF. They also had decreased risk of grade 3 or 4 toxicities compared to non-NF patients.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Survival of children and adolescents with low-risk non-rhabdomyosarcoma soft tissue sarcomas (NRSTS) treated with surgery only: an analysis of 234 patients from the Children’s Oncology Group stud...

PDF

An analysis of disease-free survival and overall survival in inflammatory breast cancer

PDF

An assessment of necrosis grading in childhood osteosarcoma: the effect of initial treatment on prognostic significance

PDF

Expansion of emergency medical service regional routing for stroke patient care in the United States from 2000-2010

PDF

Risk factors and survival outcome in childhood alveolar soft part sarcoma among patients in the Children’s Oncology Group (COG) Phase 3 study ARST0332

PDF

CAFÉ (Common Application Framework Extensible) - a framework to support data capture for clinical trials

PDF

The effect of renal function on toxicity of E7389 (eribulin) among patients with bladder cancer

PDF

The study of germ cell tumors and related conditions: an analysis of self-reported data with characterization and comparison of Family History Questionnaires respondents and non-respondents

PDF

Development and validation of survey instrument designed for cervical cancer screening in Malawi, and other low resource settings

PDF

Statistical analysis of a Phase II study of AMG 386 versus AMG 386 combined with anti-VEGF therapy in patients with advanced renal cell carcinoma

PDF

Association of oophorectomy with progression of carotid atherosclerosis

PDF

Obesity paradox in acute heart failure decompensation

PDF

The evaluation of the long-term effectiveness of zero/low fluoroscopy workflow in ablation procedures for the treatment of paroxysmal and persistent atrial fibrillation

PDF

A pilot study of chemotherapy intensification by adding vincristine, topotecan and cyclophosphamide to standard chemotherapy agents with an interval compression schedule in newly diagnosed patien...

PDF

Associations between physical activities with bone mineral density in postmenopausal women

PDF

Insulin’s effect on lactate levels in extremely low birth weight neonates. a multi-center, observational study

PDF

Lymph node metastases in non-rhabdomyosarcoma soft tissue sarcoma

PDF

Relationship of blood pressure and antihypertensive medications to cognitive change in the BVAIT, WISH, and ELITE clinical trials

PDF

Methodological approaches to assessing diurnal cortisol rythms in epidemiological studies: how many salivary samples are necessary?

PDF

Validation of the Children’s International Mucositis Evaluation Scale (ChIMES) in pediatric cancer and SCT

Asset Metadata

Creator Xia, Caihong (author)

Core Title Carboplatin and vincristine chemotherapy for progressive low grade gliomas in pediatric patients with or without neurofibromatosis type 1 (NF1)

School Keck School of Medicine

Degree Master of Science

Degree Program Applied Biostatistics and Epidemiology

Publication Date 10/20/2014

Defense Date 10/20/2014

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag chemotherapy,low grade gliomas,neurofibromatosis type 1,OAI-PMH Harvest

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Sposto, Richard (committee chair), Azen, Stanley P. (committee member), Mack, Wendy Jean (committee member)

Creator Email caihongx@usc.edu,xchwhu@hotmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-507988

Unique identifier UC11298507

Identifier etd-XiaCaihong-3022.pdf (filename),usctheses-c3-507988 (legacy record id)

Legacy Identifier etd-XiaCaihong-3022.pdf

Dmrecord 507988

Document Type Thesis

Format application/pdf (imt)

Rights Xia, Caihong

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA