[Nasional-m] Breast tumor genes signal whether cancer will kill

[Ambon]

  Breast tumor genes signal whether cancer will kill
   Gina Kolata/NYT The New York Times  Friday, December 20, 2002



NEW YORK Researchers have found a genetic signature in breast tumors that
seems to be a powerful predictor of whether the cancer will spread and kill
or whether it can easily be cured by surgery, causing no further harm.
.
The research involved relatively few patients, all relatively young, and its
conclusions remain to be confirmed by further studies. But scientists said
that the genetic signature - the activity of a collection of 70 genes -
appeared to predict cancer mortality better than traditional measures like
tumor size or stage or whether the cancer has spread to a woman's underarm
lymph nodes. In the study, 5.5 percent of women whose cancers had a good
genetic signature died within the next decade, while 45 percent of those
with bad genetic signatures died in that period.
.
It will be at least several years before the work could change medical
practice. But the findings, by researchers in Amsterdam, raise questions
about the nature of cancer, its treatment and the benefits of early
detection.
.
Doctors and patients have assumed, for example, that small tumors are more
treatable. The thinking has been that as tumors grow, they acquire mutations
that enable them to spread throughout the body. But the study, and smaller
studies that looked at genetic signatures of other cancers, indicate that
tumor size may be beside the point. Most tumors, the studies indicate,
appear to be potentially deadly, or not, from the very start.
.
Doctors have also long assumed that after surgery, to be safe, they should
treat virtually every breast cancer patient with chemotherapy, hormonal
therapy or radiation, based on the belief that it is very difficult to
differentiate women who will be helped by additional treatment from those
who do not need it.
.
The study indicates that it may soon be possible to make that distinction
and that in the future, many women will be told that chemotherapy is not
necessary. "The hope and the expectation is that we will be able to identify
subsets of patients who require less aggressive therapy, or perhaps no
therapy at all," said Dr. Todd Golub, who directs cancer genomics research
at the Whitehead/MIT Center for Genome Research in Cambridge, Massachusetts.
.
The study, published Thursday in The New England Journal of Medicine, had
its origins a few years ago when the researchers, led by Dr. Rene Bernards,
a professor of molecular carcinogenesis at the Netherlands Cancer Institute
in Amsterdam, began searching thousands of genes from breast tumors.
.
They were asking whether there was a pattern of gene activity that was
associated with a good prognosis and another associated with a bad
prognosis.
.
The methodology for such searches was developed only recently, with gene
chips. When material from tumors is washed over the chips, fragments of
active genes stick to the chips. Using fluorescent dyes to show when they
get a hit, researchers can see patterns and correlate them with a tumor's
behavior. Their initial study led the researchers to focus on 70 genes they
identified in a small study. The study indicated that the genes' activity
appeared to predict prognosis, but that needed confirmation.
.
So they began the new study involving 295 consecutive patients at the
Netherlands Cancer Institute. The women, aged 52 and younger when their
cancers were diagnosed, had had standard treatments from 1984 to 1995.
Doctors had stored tissue from the women's tumors. They knew which tumors
had spread to the lymph nodes at the time of diagnosis. They also knew
whether a woman's cancer eventually spread throughout her body or whether
she remained cancer-free after her initial treatment. And they knew who had
died of breast cancer within a decade after the diagnosis.
.
"We asked whether the gene signature would be able to predict disease
outcome in these patients," Bernards said. "The answer is very clear. It is
so powerful that it is better than any of the known criteria used today in
deciding if patients need further therapy."
.
The women whose cancer cells indicated a good prognosis had an 85.2 percent
chance of remaining free of cancer over the next decade and a 94.5 percent
of surviving the decade. Those whose cancer cells indicated a poor prognosis
had a 50.6 percent chance of remaining cancer-free and a 54.6 percent of
surviving that time.
.
Small tumors often had bad genetic signatures while large tumors often had
good ones, throwing into question a common assumption about how cancers
develop. It is often said that cancers start off unable to spread, but as
they grow, they acquire this ability.
.
"That is based on a notion that tumors go through an evolution where they
acquire more mutations as they grow bigger," Bernards said. "It says that
small tumors are inherently less aggressive and less malignant. But our gene
expression profiling shows that that notion is too simplistic, at least for
breast cancer." That is not to say that a small and harmless cancer can
never mutate to become deadly, he added, but rather that most deadly cancers
are potentially life-threatening even when they are very small.
.
Golub put it bluntly: "The metastatic potential is hard-wired at the time of
diagnosis." He said that a tumor's fate could be sealed from the first time
it was detected, no matter how small it was. He said that part of the
confusion might have come because those cancers most likely to spread also
tended to be cancers that grew quickly. So when doctors found large tumors,
they were often dangerous (the mistake may have been in assuming that when
they were small they were not). "Size per se may not matter that much,"
Golub said.
.
Golub added that many questions remained. "These genetic studies are
suggesting that your probability of metastasis may be hard-wired, but that
is different from saying metastasis has already occurred. There still may be
some importance to early detection."