Currently, once prostate cancer is diagnosed, doctors have no reliable way to know which cases are life-threatening. Doctors often have a difficult time determining whether they should monitor the cancers to see if they progress or recommend immediate treatment, such as surgery or radiation.
Both treatments can cause problems, such as incontinence and impotence.
The dilemma results in overtreatment - for example that about 48 men are treated for every life saved, says Dr Ronald DePinho of the Dana-Farber Cancer Centre in Boston.
Huge improvement in accuracy
But DePinho and colleagues have created a test they say might help doctors identify dangerous tumours more accurately than is possible now. In results published online by the journal Nature, the scientists showed an improvement in accuracy to 91% from 84%.
When applied to prostate cancer samples, the gene test indicates "how this particular cancer is wired to behave," DePinho said.
Drawing on research in mice, DePinho and colleagues identified four genes whose combined activity within cancers appears to drive prostate tumours toward being lethal. The genes are involved in processes like growth and ability to invade other tissues.
Then the researchers tested whether the combined activity of those genes also predicted cancer outcome in men, with a series of tests in human tumour samples.
In the largest test, they looked at 405 tumour specimens from men who'd been diagnosed between 1983 and 2004. Thirty-eight cases turned out to be lethal. Researchers looked for the chemical signatures of gene activity in the samples and tested how accurately that could classify tumours as lethal or not.
By itself, the gene test performed about as well as a combination of current indicators: age at diagnosis, indications of tumour spread and a "Gleason score" that assesses the appearance of tumour cells under a microscope. That standard approach was accurate 84% of the time in the study.
But accuracy rose to 91% when researchers combined that approach with the gene test. The combination "robustly predicted which men were going to die of the disease," DePinho said.
Licensing test underway
The rights to develop the test have been licensed to a company that DePinho co-founded and in which he holds a financial interest.
Experts not connected with the study praised the work but said more research on the gene test is needed.
"It's early still, but it's pretty exciting," said Dr Eric Klein of the Cleveland Clinic. "This is a step in the right direction, without question."
Dr Angelo De Marzo, a professor of pathology, oncology and urology at the Johns Hopkins School of Medicine in Baltimore, called it "extremely interesting, promising, amazing work".
It will take more research to see whether the gene test really could help doctors make treatment recommendations with more confidence than they can now, and in what situations, he said. The test might prove useful when prostate cancer is diagnosed from biopsy samples, or in deciding on further treatment after a man's prostate has been removed, he said.
De Marzo noted that some prostate cancer researchers are meeting this weekend, and "I have a feeling there's going to be a ton of buzz about this at that meeting. I think people are going to be very excited."
(Sapa, February 2011)