Biomedical Engineering Reference
In-Depth Information
Weinberg 2002). According to the clonal selection model, genomic instability in
the progressing tumor leads to multiple heterogeneous populations. Random
and sporadic combination of metastasis-enhancing properties in a few rare cells
endow them with the ability to break away from the primary tumor mass,
migrate to a distant organs, and eventually initiate metastases. Although a
possible scenario, this model appears to have difficulty explaining efficient
metastatic spread of certain small primary tumors, and the success of the so-
called ''poor-prognosis signatures,'' derived from the whole primary tumor
population, to predict metastatic risk (Bernards and Weinberg 2002).
Alternatively, mutations endowing metastatic potential could be sustained
early in tumorigenesis, giving rise to a tumor in which all cells carry a certain
metastatic propensity toward low or high malignancy. Support for early deter-
mination of malignancy comes from studies of whole tumor profiling. Risk of
metastasis and recurrence can be reliably ascertained by comparing microarray
profiles in bulk preparations of primary tumors, suggesting that most cells within
the tumor express a similarly predictive gene profile. If genetic properties impor-
tant to metastatic propensity exist in only a small subset of tumor cells, consensus
profiles could most likely not be detected. In breast cancer at least five tumor
classes have been identified by gene expression signatures that correspond to
varying prognoses for metastasis-free or overall survival. Subgroups identified
include luminal A, luminal B, normal breast-like, ERB2-overexpressing, and
basal epithelial-like (Perou et al. 1999; Sorlie et al. 2001). Overall poor prognosis
based on a panel of factors including metastasis is predicted for basal-like,
ERB2+, and luminal B as compared to luminal A and normal breast-like
subgroups (Sorlie et al. 2001). Bulk profiling has also been shown to predict
prognosis based specifically on metastasis (van't Veer et al. 2002; van de Vijver
et al. 2002; Ramaswamy et al. 2003; Wang et al. 2005) and the profiles of distant
metastatic cells have been shown to closely resemble that of the primary tumor
(Weigelt et al. 2005a). If these profiles could be further subdivided based on
specific oncogenic transformation they could become powerful tools both in
research connecting metastatic potential to particular mutations and in develop-
ing more focused diagnosis and treatment. In fact in mouse models of various
breast tumor types, profiling has revealed a generic tumor signature as well as a
specific signature dependent upon initial oncogenic events associated with indi-
vidual prognoses (Desai et al. 2002). While further work is needed to extend this
finding to human tumors, it offers further evidence for early establishment of
tumor character with regard to metastasis.
Tumor-wide expression analysis still does not address the question as to how
mutations are maintained throughout potentially years of tumor growth. One
way to reconcile metastatic character and early determination is the idea of
adaptation. Metastatic potential could arise from a somatic mutation that
allows the cell to respond in a non-cell-autonomous fashion to signals in the
microenvironment. For example, later in tumor progression, stroma ''acti-
vated'' by the presence of a tumor (through inflammatory pathways for
instance) could provide signals stimulating the epithelial to mesenchymal
