|
||||||||
|
DISEASE PROFILING Gene Expression Pattern Predicts Potential for Tumor MetastasisSome tumors may be born to spread. Researchers have found a gene expression pattern that distinguishes primary tumors from those that have metastasized in several types of cancer, but surprisingly, some primary tumors also had a metastatic gene expression pattern and a poorer prognosis. The DNA microarray study appears in the Dec. 9 online Nature Genetics.
At an early stage, the bulk of a tumor--and not just a few renegade cells--may be poised to spread from a primary location to other tissues in the body, according to a DNA microarray analysis of 279 samples of several different tumor types by (l to r) Sridhar Ramaswamy, Todd Golub, and their colleagues at the Whitehead/MIT Center for Genome Research Eric Lander and Ken Ross. (Photos by Steve Gilbert. Lander photo by Sam Ogden) "These findings support the emerging notion that the clinical outcome of individuals with cancer can be predicted using the gene-expression profiles of primary tumors at the time of diagnosis," write the authors, led by oncologist Sridhar Ramaswamy, HMS instructor in medicine at the Dana-Farber Cancer Institute and a researcher at MIT's Whitehead Institute. The results run counter to a widely accepted view of cancer progression and metastasis, in which tumors acquire a series of genetic mutations allowing them to grow, dominate, and eventually spread. This summer, other researchers cited the unpublished data as evidence that "some cancers start out on the wrong foot," and metastasis may be preordained by early mutations.
"This is one of the first times that anybody has taken databases of different populations, looked for features in data that cut across different kinds of cancer, and used them to explain an important aspect of tumor biology, in this case a gene signature related to metastasis," said Paul Meltzer, senior investigator in the cancer genetics branch of the National Human Genome Research Institute. "This points the way to the future." Telltale GenesIn this study, the researchers first analyzed gene-expression profiles of 12 metastatic nodules of cancers with origins in the lung, breast, prostate, colon, uterus, and ovary. They compared the profiles to those from 64 primary cancers from the same sites of origin from different individuals. Using a computer algorithm, they found 128 genes turned on or off differently in metastatic samples compared with primary tumors, no matter what the tissue of origin.Oddly, the researchers found some primary tumors expressing the metastatic pattern. Because the study was done on whole tumors, most cells of the samples would have to express these genes to be detected using DNA microarrays. "That observation made us rethink the prevailing notion that metastases arise from a few breakaway cells," Ramaswamy said. "It made us think that metastatic potential is programmed into the bulk of primary tumor cells." Following that train of thought, the researchers analyzed gene expression samples of primary tumors in several other databases. In 62 samples of early-stage primary lung tumors, the 128 metastasis-associated genes separated these tumors into two groups. The researchers found significantly shorter survival times overall in people whose primary tumors were expressing the metastasis-associated profile. The researchers also were able to narrow the 128-gene set to only 17 genes that seemed to contribute most to the distinction. The refined set also significantly predicted the metastatic outcome in a dataset of 78 small early-stage primary breast cancers and another database of 21 prostate adenocarcinomas. The gene expression pattern of the smaller set also correlated with patient outcome in a group of 60 medulloblastomas. The signature did not predict outcome in diffuse large B cell lymphoma which, like other cancers of the circulatory system, is believed to have very different mechanisms of spreading. In their study, the researchers benefited from the Human Genome Project and a culture of openness. They used databases accessible on the Web from other labs and other countries. They also were able to analyze and compare profile patterns across different microarray designs. "This supports the contention that the technology may be ready for prime time in the not-too-distant future," said oncologist Carlos Caldas, a professor at the University of Cambridge and Addenbrooke's Hospital in England and senior investigator of its cancer genomics program. "We're at least two to three years away from having evidence convincing enough for clinical use. But this shows the way we're going in terms of eventually individualizing prognosis, individualizing treatment, and, hopefully, identifying new molecular targets and developing new therapies. Cancer is a disease first of the genome and ultimately of the proteome." Caldas predicts further refinements will come from analyzing tumor DNA and proteins. Uncovering Gene OriginsAlong with the enthusiasm comes a widely acknowledged cluster of caveats and concerns. Perhaps the biggest one is that researchers are not exactly sure which cells are expressing which genes. Tumors contain many different cell types. In this study, researchers suspect the gene set associated with metastasis in primary tumors could represent either a kind of monologue of gene activity coming from tumor cells or a dialogue between the tumor and surrounding stromal tissue. They are working on identifying the cellular origin of individual genes in the set."The fact that you can make any statement at all about the biology of such diverse tumors is interesting," Meltzer said. "Microarrays provide a read-out of the behavior of a complicated network of genes. The challenge for the future is to understand mechanistically how this is related to the biology of the tumor." And that's just the microarray analysis. Metastasis is a complex phenotype with many distinct biological behaviors coordinated by many genes at different times, allowing a cell to move from the confines of its original tissue to another discrete part of the body. In this study, some of the metastatic gene signature seemed to come from nonepithelial cells, perhaps from surrounding normal stromal cells, which have been implicated in tumor growth. "I would be the first person to say that these findings are not yet ready for clinical application," Ramaswamy said. "The signature has predictive power in a population. That's different from saying we can diagnose individual patients accurately. I wouldn't even want to term this the end-all and be-all signature. It may be that the core genes we're dealing with change slightly with studies of larger tumor sets. But we've seen a glimpse into something that may be applicable to patients, after we cross a lot of hurdles." --Carol Cruzan Morton |
