The standard concept of how cancer starts is that malignant tumors arise from a single cell transformed by a chemical carcinogen, oncogenic virus, radiation damage, endogenous genetic damage caused by oxidative insult to DNA, or any of a host of other potential ways (e.g., chronic infections with a bacteria such as H. pylori or with a parasite such as schistosomiasis, or hormonal imbalance). Once the initiated cell starts to undergo clonal expansion, it undergoes multiple genetic changes, due to genetic instability,
leading to an invasive metastatic cancer. This progression is thought to occur sequentially,
as exemplified by the work of Vogelstein and colleagues on colon cancer.46 The idea here
is that colon cancer goes through a series of ‘‘evolutionary’’ changes from hyperplasia, to earlystage
adenoma, to late-stage adenoma, to carcinoma, and finally to metastatic cancer.
There is, however, another point of view proposed by Weinberg and colleagues.56,57 This hypothesis, for which there are supportive clinical data, states that the genes involved in driving invasiveness and metastasis may be expressed early in the progression pathway and actually be the same genes involved in a selective growth advantage for these cells. These cells may be lurking even in early-stage cancers. That is, some cancers are predestined almost from the beginning to evolve into invasive, metastatic
tumors and some are not. This possibility has huge implications for cancer screening, diagnosis,
and choice of therapy. Numerous women receive a diagnosis of ductal carcinoma in situ of the breast based on mammography screening, and many men receive a diagnosis of prostate cancer based on a prostate-specific antigen (PSA) test and subsequent biopsy. And yet many of these patients have indolent tumors that would not affect their overall life expectancy, and they still often undergo significant surgical
and drug treatments. The problem is that we are only beginning to be able to tell (e.g., by gene
expression arrays) which of these so-called early-stage cancers will be lethal and which ones won’t.
Another point of the Weinberg theory is that the genetic alterations that occur during tumor progression do not necessarily occur in a given sequence and are probably different for different cancers.56 One might even suggest that they may be different in different patients who have the same histological tumor type. Ultimately, however, these genetic and phenotypic changes lead to a similar loss of cell proliferation control and expression of a panoply of genes (maybe not the identical ones) that make some tumors invasive and metastatic. There are clinical data supporting some of these concepts. In a study by van de Vijver et al.,58 it was determined that the gene expression profile of breast cancers was a much better predictor of disease outcome in patientswith breast cancer than standard clinical and histopathological staging. Indeed, they could restratify patients listed as low risk or high risk by clinical staging into a more accurate prognostic outcome category (based on actual metastasisfree survival) through gene expression arrays. In addition, Al-Hajj et al.59 were able to identify and isolate the more tumorigenic cells from a
heterogeneous population of breast tumors in eight of nine patients. These more aggressive cell types were identified by their cell surface markers and by repeated passage in nude mice. Each time the more aggressive cells were injected into nude mice they produced tumors, whereas the marker-negative cells did not grow. These data suggest that the aggressive tumorigenic cells can be prospectively identified in initial tumor biopsies containing mixed populations of cells and can be used to discriminate patients
with potentially more aggressive tumors.
