Trial Information

A Phase II Trial of Poly-ICLC in the Management of Recurrent Pediatric Low Grade Gliomas

Background/Rationale The incidence of primary pediatric brain tumors in the United States is
about 1500 per year. Brain tumors are the most common solid tumor diagnosed in childhood and
thus account for significant childhood mortality in the United States. Low-grade
astrocytomas and gliomas are the most common type of brain tumor of childhood (36% of
childhood brain tumors). These tumors encompass a heterogeneous assortment of histological
subtypes including: fibrillary, protoplasmic, gemistocytic, and mixed variants. Pilocytic
astrocytomas, pleomorphic xanthoastrocytomas and subependymal giant cell astrocytomas are
also included. Furthermore, in young children there are some unique rare entities that
behave like low-grade tumors, including infantile desmoplastic gangliogliomas, and
desmoplastic astrocytomas. Although children with low-grade astrocytomas often survive many
years after conventional treatment with surgery and sometimes radiotherapy, some children
will not fare as well. These tumors constitute a heterogeneous group because of differing
locations within the brain and varying biological behavior of different subtypes. For those
where total excision is possible, such as cerebellar astrocytomas, prognosis is excellent
with over 90% ten-year survival rates with surgical excision alone. In contrast, survival
rates in children with cerebral or diencephalic tumors are 40-70% at five years with
irradiation, but decline to 11-50% at 10 years (Mundigers, 1990). Some tumors however may be
unresectable/partially resectable, and radiation can have undesirable side effects in young
children. While the most significant intellectual deficits occur in young children less than
5 years treated with cranial irradiation, the deficits recognized even in young adults
warrant extending the age to 10 years for avoiding radiation. Chemotherapy regimens are used
for high-risk patients (progressive tumor, residual tumor) as a means to avoid or delay
radiation in young patients, but side effects of chemotherapy are frequently reported.

Newer forms of effective treatment that will have lesser side effects are much needed in
childhood brain tumors especially low-grade gliomas. We propose to study the efficacy and
toxicity of poly-ICLC, a biological response modifier in children with low-grade gliomas.

PROTEOMICS

Current diagnostic and therapeutic monitoring of brain tumor patients are significantly
hindered due to limited understanding of brain tumor biology and response to therapy. The
majority of CNS tumors cannot be identified or followed by expression of serum or CSF
markers. However, if available, such markers would be highly desirable and could be used to:

- Detect minimal residual disease

- Predict response to specific targeted therapies

- Predict or anticipate tumor progression

- Distinguish tumor recurrence from post surgical changes or post-radiation changes on
neuro-imaging

- Augment current histopathologic classification systems

- Improve current clinical and pathological treatment stratification schemata

- Assess efficacy of and tumor response to specific biologic targeted therapies that may
not impact tumor size as a primary tumor endpoint (e.g., small molecule inhibitors or
anti-angiogenic strategies) While such markers would be useful to prognosticate,
monitor and treat all CNS tumors, its use in glial tumors including recurrent low grade
astrocytomas is critical since these tumors are often biopsied at presentation, but not
at recurrence. Often these tumors are not amenable complete resection or biopsy due to
the eloquence of brain tissue they infiltrate (e.g., optic pathway, brainstem or
hypothalamic gliomas), or the blood vessels that they encase.

CNS biologic material in CSF Glial tumors tend to disseminate locally along white matter
tracts rather than through sub-arachnoid seeding. Dissemination of low grade gliomas along
the sub-arachnoid space has been reported in children with low grade gliomas. Even focal
tumors are frequently adjacent to CSF pathways (e.g., intrapeduncular fossa, third and
fourth ventricles) resulting in direct contact between tumor tissue and spinal fluid. Yet
examination of CSF cytology for these tumors is not standard. Given limitations of
identifying tumor cells in the CSF, methodologies that could improve our understanding of
CNS tumors of all types are needed. This would provide a significant improvement in
currently available knowledge about the biology of these tumors, and could elucidate
potential therapeutic avenues.

Proteomics, a relatively new area of research whereby total protein complement of a tissue
compartment is analyzed, has successfully been used to identify novel biomarkers in solid
tumors (Zheng, 2003);( Khwaja, 2007). Because proteins are effectors of all cellular
functions, their measurement should represent the most direct means of cellular
characterization and hence tumor biology. Because cells and their environment exist in an
integrated state, it has been possible to interrogate the proteins of extra-cellular
compartments to assess the presence and impact of tumor cells. This has been done primarily
using serum or plasma to establish a method of screening for the presence of low stage
tumors. An analogous extra-cellular compartment for use in brain tumors would be
cerebrospinal fluid (CSF). It circulates throughout the CNS and exchanges proteins with the
extra-cellular fluid of the brain and spinal cord.

CSF is continuously created and reabsorbed, providing a real time steady state proteome.
Unlike serum, which contains a highly complex protein mixture ranging from very low
abundance proteins in the 10-30 pg/mL range to very abundant proteins in the 35-55 mg/mL
range, CSF contains a less complex protein mixture (Omenn, 2005).

Therefore, the CSF is more likely to contain higher relative concentrations of
tumor-specific proteins (higher signal to noise ratio) than serum. Taken together this makes
CSF and attractive alternative to serum for detection of brain tumor related biomarkers.
Unlike leukemia and many solid tumors outside the CNS, where serial biopsies are readily
performed, tumors of the CNS are not easily accessible other than at the time of initial or
repeat resection or biopsy. While studies on these samples provide important findings
regarding tumor biology, serial analyses during treatment are not reasonable. By contrast,
the CSF of tumor patients can be more readily sampled in most pediatric patients. With the
development of proteomic technology, investigation of tumor related signals at the time of
diagnosis through treatment, and then in remission and/or at the time of recurrence or
progression is possible.

While CSF for seeding tumors is readily available and routinely obtained for cytology, the
systematic evaluation of the proteins within these samples could be of considerable
scientific importance. In addition to identifying potential makers of disease or response to
therapy, the glycosylation and phosphorylation status of many proteins can also be
evaluated. Studies in tumor tissue show that such information reveals activity of different
enzymes that correlate with treatment response (Mellinghoff, 2005); (Helgi, 2005) or
progression of leptomeningeal metastases (Brandsma, 2006).

Proteomics CSF proteomics has been applied to many neurological disorders including
Alzheimer's disease, amyotrophic lateral sclerosis, multiple sclerosis, acute brain injury
and Creutzfeldt-Jakob disease (Rohlff, 2001).

Reports of its use in neuro-oncology are limited, but demonstrate the potential of this
technology to effectively identify tumor biomarkers. One study used two dimensional
polyacrylamide (2-D) gel electrophoresis to measure the relative quantities of two
pre-selected markers, N-Myc and l-CaD, in the CSF of brain tumor patients (Zheng, 2003).
Another used ELISA of CSF to identify Osteopontin as predictive of AT/RT and correlated with
response to therapy (Kao, 2005). CSF proteomics using 2-D gel electrophoresis in combination
with mass spectroscopy and cleavable isotope Coded Affinity Tag (cICAT) was used to evaluate
60 samples of CSF and tumor cyst fluid taken from adults with brain tumors and
non-neoplastic controls. These techniques were used to find a panel of proteins
differentially expressed in lower vs. higher-grade gliomas. Findings were confirmed using
Western Blot analysis probing for eight selected proteins based on implied role in
gliomagenesis and availability of antibodies. This report, which has been accepted for
publication pending revisions, identified 21 potential CSF biomarkers for astrocytoma.

As mentioned above, there is evidence that gliomas disseminate through the subarachnoid
space. Currently there are several consortia actively studying protein expression in the
spinal fluid of children with malignant glial and embryonal tumors (Pediatric Brain Tumor
Consortium, Pediatric Oncology Experimental Therapeutics Consortium). Proteins interrogated
in these protocols include those involved with angiogenesis and neovascularity (EGF, VEGF
and bFGF), those involved in tumor growth and migration (Secreted protein with acidic and
cysteine rich domains (SPARC), attractin). There is no consortium actively collecting spinal
fluid sample in children with lower grade tumors. A secondary goal of this study is to
examine these proteins in the CSF of children with low grade gliomas who have tumor
progression. Comparison of CSF protein expression of in high grade and low grade tumors is
likely to help identify biological markers specific for tumor progression, or for tumor
pathology.