There’s a revolution brewing in the diagnosis of cancer that could dramatically change how doctors figure out which tumors are truly life-threatening – and need chemotherapy — and which are not.
In the Netherlands, the new tool – called by various names, including gene expression profiling – is expected to be available for some women with breast cancer as soon as May. In the US, it could be several years before the technique is routinely available.But if gene profiling lives up to its promise, thousands of patients a year with tumors deemed unlikely to spread might safely skip chemotherapy. At the other extreme, people with cancers deemed lethal might skip chemotherapy if it’s unlikely to work, and go straight to experimental or alternative therapies.
With early stage breast cancer, for instance, doctors currently vastly over-treat because they can’t tell which patients need chemotherapy and which don’t. Dr. Eric Winer, director of the breast oncology center at the Dana-Farber Cancer Institute in Boston estimates that in some situations, 100 women get chemotherapy for every 20 helped and in other cases, it’s 100 to 1.
With gene profiling – also known as genetic fingerprinting, molecular signatures and molecular profiling – the idea is to determine which genes are expressed, or turned on, in a given tumor.
Once genetically analyzed, tumors can then be categorized as having a good prognosis or not, and treatment can be chosen accordingly. Gene profiling will likely be used in addition to measures doctors already use, such as tumor size and whether the cancer has spread to lymph nodes. It will also be used to determine whether a tumor is likely to respond better to drug A or drug B.
Various teams of researchers have studied gene profiling in a number of different cancers and so far, perhaps surprisingly, there is very little overlap in the genetic patterns.
Several years ago, Charles Perou, now assistant professor of genetics at the University of North Carolina, with David Botstein and Pat Brown of Stanford University, identified a set of 450 genes that can predict outcome in breast cancer.
At Dana-Farber Cancer Institute in Boston, Dr. Sridhar Ramaswamy and his team have identified a set of 17 genes that seem to be harbingers of metastasis in various cancers. And in the Netherlands, Dr. Marc J. van de Vijver, Rene Bernards and others have identified a set of 70 genes that can predict which breast tumors will metastasize, or spread to other parts of the body, and which won’t.
Researchers have also used gene expression to predict outcome in cancers of the kidney, lung, and prostate. In B-cell lymphomas, they have identified one set of 16 genes and another of 13 genes that can do likewise. In one subtype of ALL (acute lymphocytic leukemia), researchers found a set of 20 genes that can predict which patients will develop a secondary cancer called AML (acute myelogenous leukemia).
This dazzling, even frightening, ability to predict the future is possible because of commercially available “gene chips” or DNA micro-arrays.
Imagine a thin, roughly one inch square plate of glass containing thousands of tiny spots of DNA, each spot representing a single human gene. Researchers then take genetic material called mRNA from a tumor and spread it on the DNA. If the mRNA finds its match on a spot of DNA, it sticks, and, because it’s been pre-treated, fluoresces under a special light. This technique – at roughly $1000 a test – tells researchers which genes are active in that particular tumor.
The idea that a tumor’s potential to metastasize may be knowable while a tumor is still very small is both encouraging and distressing.
Historically, doctors believed tumors started off benign and mutated to acquire the ability to metastasize. ”Our study shows that this notion, at least for breast cancer, is wrong,” says Bernards, the senior author of the Dutch study on 70 predictive genes and head of the division of molecular carcinogenesis at the Netherlands Cancer Institute.
In other words, ”small does not mean benign,” he continues. ”We have seen small tumors that are poor in outcome and we have seen large tumors that have a good prognosis.”
That concerns Dr. Larry Norton, head of solid tumor oncology at Memorial Sloan-Kettering Cancer Institute in New York. ”It’s a big stretch to say if a cancer is bad, you might as well give up,” he says. ”It may be that those tumors are the most curable with therapy.”
Dr. Steven Goodman, a biostatistician and epidemiologist at the Kimmel Cancer Center at Johns Hopkins University is also cautious. So far, he says, ”there is absolutely zero evidence that by deferring chemotherapy based on these predictions you are better off.”
Despite concerns by some scientists that large prospective clinical trials should be conducted before gene expression profiling hits the clinic, the technique appears be on the fast track.
Bernards, who collaborated with scientists from Rosetta Inpharmatics (a wholly-owned subsidiary of Merck & Co.) on the study of 70 predictive genes, now has his own company, Agendia, Inc. in Amsterdam that, he says, ”allows us to offer this test.”
Celera Diagnostics is also doing a study correlating gene expression in breast tumors with progression of disease, with results expected later this year, says Tom White, chief scientific officer.
At Massachusetts General Hospital, a team led by Dr. Daniel Haber, director of the Center for Cancer Risk Analysis, with colleagues from Brigham and Women’s Hospital and Dana-Farber, has started testing 500 frozen breast cancer samples, in collaboration with the Dutch group. Funded by the Avon Foundation, the team hopes to test 5,000 samples in the next few years. ”Clearly, this is at the cutting edge of oncology,” says Haber, though he warns that it should be validated in a large prospective study before patients use it to make decisions about treatment.
One development that could speed things up is an idea being tested independently at Celera Diagnostics, Genomic Health, Inc. and researchers from Duke University and the University of North Carolina.
Instead of using frozen samples of tumor tissues, as the Dutch do and others do, these scientists are testing genetic profiling on tumor tissue that has been ”fixed” in formalin (formaldehyde) and embedded in paraffin wax for long term storage.
Until recently, scientists had thought formalin-fixed tissue “was unstudy-able,” says Dr. Matthew Ellis, clinical director of Duke’s breast cancer program, because the RNA in the tumors degraded. “That turns out not to be the case,” he says. ” We can do profiling on formalin-fixed tissue, and that is incredibly powerful.”
If this strategy works, “we can analyze 10 years’ worth of information in six months,” says Ellis.