225generate a dose-response curve for each chemotherapeutic agent tested on a given
patient specimen. Features of each dose-response curve are used to score a tumor’s
response to each ex vivo treatment as “responsive,” “intermediate response,” or
“non- responsive.” Collectively, these scores are used to assist an oncologist in mak-
ing treatment decisions.
Genomic Approaches to Predict Response to Anticancer Agents
Gene Expression Patterns to Predict Response of Cancer to Therapy
Human lymphoblastoid cells, immortalized white blood cell lines derived from dif-
ferent healthy individuals, display considerable variation in their transcription pro-
fi les, which underlies interindividual susceptibility to DNA damaging agents. Gene
expression, measured by Affymetrix GeneChip Human Genome U133 Plus 2.0, has
been associated with sensitivity and resistance to DNA-damaging anticancer agents
(Fry et al. 2008 ). A cell line from one person would be killed dramatically, while
that from another person can be resistant to exposure to the anticancer agent. Using
computational models to pinpoint differentially expressed genes with positive or
negative correlations, the investigators identifi ed 48 genes whose pre-treatment
expression could predict sensitivity to anticancer agent MNNG with 94 % accuracy.
MNNG alkylates certain DNA bases, leading to mutagenesis. Some of this damage
can be repaired by the DNA methyltransferase MGMT. But if it is not, the DNA
mismatch repair or MMR pathway targets damaged DNA bases and sets off apop-
tosis. Consequently, cells with reduced MGMT activity but a functional MMR path-
way are expected to be more sensitive to MNNG, whereas cells defi cient in both
pathways are more MNNG resistant but accumulate mutations when exposed to the
compound. Because gene expression is the most accurate predictor of alkylation
sensitivity, there are good prospects for translating these fi ndings to a clinical set-
ting to predict whether a tumor will respond to alkylation chemotherapy.
Genomic Analysis of Tumor Biopsies
Genomic Health Inc is developing a service to provide individualized genomic anal-
ysis of tumor biopsies to physicians as a guide to treatment of patients with cancer.
Fixed paraffi n-embedded tissues (FPET), stored tumor tissue samples collected
over the past 20 years, are used for this purpose. Instead of waiting years to accu-
mulate fresh tissue and track patient outcomes, Genomic Health’s FPET analysis
can be performed using routinely stored biopsies from patients with known out-
comes therefore accelerating clinical trials. RNA analysis of thin sections of stan-
dard tumor biopsies is used to evaluate panels of genes that may predict breast
cancer recurrence and response to chemotherapy as well as response to EGFR
inhibitor therapy in lung cancer. This approach is now being tested in clinical trials
Determination of Response to Therapy