2012 EDUCATIONAL BOOK - American Society of Clinical Oncology

next 2 to 3 years. In addition to bevacizumab, several other
agents targeting vascular endothelial growth factor (VEGF)
or its receptor have been tested in clinical trials. In general,
reported radiographic response rates and PFS rates have
been similar or lower with these agents compared with
bevacizumab. 17,18 In one recent phase III study testing the
VEGF receptor inhibitor cediranib as monotherapy or with
lomustine compared with lomustine alone in recurrent GBM
found no significant difference in median PFS between these
groups. 19
The Problem of Bevacizumab Failure
Although bevacizumab and other antiangiogenesis agents
have demonstrated improved radiographic response rates
and PFS rates in recurrent GBM compared to prior cytotoxic
and targeted therapies, it is less clear that these agents are
really improving OS. In addition, multiple studies to date
have failed to demonstrate a benefit of combination therapy
with bevacizumab. In the setting of bevacizumab failure,
studies of various salvage therapies have demonstrated low
radiographic response rates and 6-month PFS rates in the
0% to 2% range. 20,21 This dismal prognosis indicates that
progression in the setting of antiangiogenic therapy likely
represents a change in biology of the tumor. However, the
mechanisms of resistance and strategies for optimizing
treatment at this stage of disease remain an active area of
investigation. Because of concern regarding rebound edema
with discontinuation of bevacizumab, many practitioners
tend to continue bevacizumab in this setting despite radiographic
progression. Although data related to the question
of continuation or discontinuation of bevacizumab at the
time of bevacizumab progression continues to evolve, a
recent single institution meta-analysis suggested an improved
outcome with continuation of bevacizumab in this
situation. 22
Opportunities and Barriers for Improved
Treatment of GBM
Molecular subtypes and candidate therapeutic targets in
GBM. One of the recent therapeutic themes in oncology is
that the identification of signature molecular alterations in
particular tumors in combination with drugs that potently
and specifically target that alteration can have dramatic
effects on patient outcome. The term “oncogene addiction”
describes the hypothesis that tumor growth and survival is
dependent on activity of one key gene or pathway 23 and can
be seen in successes of targeted therapies in specific tumor
subtypes such as imatinib for BCR-ABL-positive CML or
BRAF inhibitors for metastatic melanoma with BRAF mutations.
Although similar dramatic advances in GBM have
not been observed yet, notable increases in our knowledge of
the molecular underpinnings of GBM are raising hopes for
targeted breakthroughs in therapy.
Strong data indicate that the pathologic entity defined
under the single WHO diagnosis of GBM consists of multiple
molecular subtypes of tumors. These tumors all share the
histopathologic features of GBM, but demonstrate very
distinct biology and molecular ontology. These data may
point the way to more effective or “personalized” treatment
based on the molecular features of an individual patient’s
tumor. 24 In particular, several studies have identified gene
expression differences between GBM subtypes. One prominent
subtype, referred to as “mesenchymal,” is characterized
110
by increased levels of expression of genes associated with
mesenchymal differentiation, extracellular matrix, invasion,
and angiogenesis. These “mesenchymal” tumors are associated
with worse prognosis and may be associated with
“primary” GBMs. Another robust gene expression GBM
subtype identified and validated in multiple data sets is
characterized by increased expression of genes associated
with normal neural tissues or neural development, and has
been called “proneural.” These proneural tumors are associated
with better prognosis and “secondary” GBMs. 25-27
The recent integrated molecular analysis from The Cancer
Genome Atlas Network (TCGA) effort in GBM further found
that specific somatic mutation and DNA copy number alterations
were associated with specific gene expression subtypes.
27 Proneural tumors were associated with significantly
higher rates of point mutations in IDH1 (p � 0.01) or p53
genes (p � 0.1), and amplification of PDGRA (p � 0.01). In
contrast, mesenchymal tumors were characterized by higher
rates of chromosomal deletions involving NF1 or mutations
in NF1. Another gene expression subtype called Classical
was associated with higher rates of amplification of
chromosome 7 (including amplification of EGFR) and loss
of chromosome 10. Since these molecular alterations are
associated with differences in clinical behavior and prognosis,
these and other less-dominant molecular alterations
are potential candidates for therapies targeting particular
tumor subtypes.
Predictive and Prognostic Biomarkers in GBM
HOWARD COLMAN
In addition to a better understanding of the molecular
pathogenesis of gliomas, the growing molecular profiling
data have led to the identification of biomarkers that are
beginning to be used to help guide patient therapy. Clinically
relevant biomarkers can be roughly divided into those
that are prognostic and predictive. Prognostic markers are
associated with patient outcome or natural history of the
disease in patients without treatment or in patients receiving
nontargeted therapies. Predictive biomarkers are associated
with differential outcome to treatment with a specific
targeted therapy. Recent studies have identified several
molecular markers prognostic of patient outcome to standard
therapy (radiation and temozolomide) in GBM. These
include MGMT promoter methylation 28 and a multigene
predictor based on gene expression levels of multiple
genes. 25 Mutation in the genes IDH1 and IDH2 was recently
described in low- and high-grade gliomas. 29 Mutation of
IDH1 or IDH2 is observed in a small subset of GBM tumors
and is strongly associated with the proneural gene expression
phenotype and better prognosis. In addition, evidence
from TCGA has also identified an epigenetically defined
tumor subset that overlaps with IDH1 mutation. This
CIMP, is analogous to prior descriptions of CIMP in colon
cancer and other tumors, although the identity of the genes
methylated in GBM is distinct from these other tumors. 30
CIMP-positive tumors are positively associated with IDH1
mutation and show improved prognosis, compared to CIMPnegative
tumors. Future work is required to define the
causative role of CIMP, IDH mutation, and their relation to
gene expression phenotype and gliomagenesis in GBM and
lower grade gliomas.