Prostate tumors with a distinctive four-gene “signature” are far more lethal than others, laying the groundwork for a test to predict which tumors need aggressive treatment, U.S. researchers said on Wednesday.

Tests of this four-gene signature method alone accurately identified 83 percent of deadly prostate tumors from tissue samples taken in a national health study.

When they combined this method with a standard test of a prostate tumor’s aggressiveness, the team accurately identified more than 90 percent of tumors that later killed patients.

“This would have 92 percent accuracy relative to what we currently have, which is at best 75 percent accuracy,” said Dr. Ronald DePinho of Dana-Farber’s Belfer Institute for Applied Cancer Science in Massachusetts.

“There is no question this will influence the practical management of these cases,” DePinho, whose study appears in the journal Nature, said in telephone interview.

He said such a test would spare many men from unnecessary treatment for cancers that might never have killed them.

“The vast majority of prostate cancers would not become life-threatening, even if left untreated. But because we can’t accurately forecast which are likely to spread and which aren’t, there is a tendency to unnecessarily subject many men to draconian interventions,” he said in a statement.

Currently, he said, about 48 men must be treated for prostate cancer to save one life, and the main forms of prostate cancer treatment - surgery and radiation therapy - can cause impotence and incontinence.

“This will clearly shift the numbers of individuals that are treated,” DePinho said.

AGGRESSIVENESS GENES

To find genes that drive the aggressiveness of prostate tumors, DePinho said, his team “ping-ponged” between mouse and human studies.

They started out with mice that lack a working copy of the Pten gene, which is involved with cell growth. These mice develop tumors, but the tumors do not spread.

They looked to see what genes kept those tumors from spreading and found a gene known as Smad4 that acts as a brake to cancer growth.

Blocking this gene in mice that also lacked a working copy of Pten appeared to flip a growth switch. Mice without working copies of both genes developed fast-growing tumors that spread to their lymph nodes and beyond.

Once they had narrowed down what kept the cancers from growing, DePinho’s team conducted a series of experiments to find genes that were key to driving this growth. They found two: SPP1 and CyclinD1. Both genes play a role in cell division and both work closely with Smad4.