Biomedical Engineering Reference
In-Depth Information
Along similar lines, a study by Stull
et al
. performed gene expression microarray
and bioinformatics analysis using xenograt tumor samples generated by injecting
human cancer cell lines derived from dif erent organs to identify common cell
surface molecules (including CAMs) as therapeutic targets (Stull
et al
. 2005).
h eir analytical approaches included principal component analysis (PCA) and
multivariate analysis of variance (MANOVA) on all tumor data and on individual
tumors. Using PCA, it is possible to identify genes that account for the majority
of the gene expression variance across tumor types. In contrast, MANOVA can
be used to identify combinations of genes that are most dif erentially expressed
between tumor types. By isolating genes that contribute both to high variance
across tumor types (using PCA) and dif erential expression between tumor types
(using MANOVA), the authors identii ed 12 informative genes (Stull
et al
. 2005).
IDENTIFICATION OF MULTIPLE TUMOR-SPECIFIC CAMs
AS TARGETS FOR MULTIMERIC LIGANDS
h e methods discussed earlier were focused on identifying one or more CAMs
involved in tumor progression and metastasis. However, they were all focused
on targeting single tumor- or organ-specii c CAMs using monomeric ligands
such as phage-display peptides or antibodies. It is well known that tumors are
heterogeneous. It is dii cult to i nd an individual CAM expressed in every cell of
an individual patient tumor or all the patients suf ering from same tumor. To cast
a wider net for CAMS with therapeutic potential, Balagurunathan
et al
. (2008)
searched for CAMS that could be targets of multimeric ligands by comparing 28
dif erent unique tissues/cell types to pancreatic tumors. h ey chose to study cell
surface molecules by restricting their analysis to cell surface molecules present on
the Agilent Human 1A array. As a way of focusing on post-translational targets,
the authors combined stringent gene expression criteria with tissue microarrays at
the protein level. To identify potentially high avidity, tumor-specii c CAMs, their
algorithm screens for combinations of receptors that appear together in tumors
but are either absent or only partially represented in normal tissues. Based on
binding avidity arguments, they estimated that drug target sites with therapeutic
potential could be identii ed as those heteromultimeric binding sites that express
at least two additional CAM receptors in pancreatic tumor tissue relative to normal
tissue. Furthermore, they demonstrated by tissue microarrays that in their study
the target combinations of cell surface molecules covered 82% of patient samples.
Hence they demonstrate that multiple CAM targets can be identii ed using gene
expression and bioinformatics and, further, suggest that multiple combinations of
CAM targets will be needed to treat the broadest possible spectrum of pancreatic
cancer patients (Balagurunathan
et al
. 2008).
