Scientists find blood signatures for aggressive prostate cancer

* Studies find distinct genetic patterns in aggressive cases

* "Barcode" blood test can read genetic changes

* Prostate cancer is second most common cancer in men

* Key to best treatment is finding life-threatening tumours

By Kate Kelland

LONDON, Oct 9 (Reuters) - Scientists have found two distinctgenetic "signatures" for prostate cancer that may help doctorspredict which patients have aggressive tumours, and designedexperimental blood tests to read those genetic signs likebarcodes.

The teams, whose work was published on Tuesday in the LancetOncology journal, believe tests developed from the signaturescould eventually be used to tell which patients need immediatetreatment.

"Prostate cancer is a very diverse disease - some peoplelive with it for years without symptoms but for others it can beaggressive and life-threatening," said Johann de Bono, who led astudy at Britain's Institute of Cancer Research. "So it's vitalwe develop reliable tests to tell the different types apart."

Researchers in Britain and the United States found that byreading the patterns of genes switched on and off in bloodcells, they could accurately detect which advanced prostatecancer patients had the worst survival rates.

Prostate cancer is the second most common cancer in menafter lung cancer. There were 899,000 new cases diagnosedworldwide in 2008, the last year for which there is full globaldata, according to the World Health Organisation's InternationalAgency for Research on Cancer.

While many cases can progress quickly, spreading to otherorgans and becoming deadly, experts say as many as half ofprostate cancers are likely to remain confined to the prostateand are unlikely to become life-threatening.

The problem has always been knowing accurately, and at anearly stage, which tumours are most likely to kill.

Although tests for aggressive forms of prostate canceralready exist, experts say they are only moderately accurate.

De Bono said scientists can learn more about prostatecancers by the signs they leave in blood. This allowed his teamto develop a test potentially more accurate than those availablenow and easier for patients than taking a biopsy, he said.

"Our test reads the pattern of genetic activity like abarcode, picking up signs that a patient is likely to have amore aggressive cancer. Doctors should then be able to adjustthe treatment they give accordingly," he said in a statement.


For his study, De Bono's team scanned all the genes in bloodsamples from 100 patients in London and Glasgow with prostatecancer. They included some already diagnosed with advancedcancer and some thought to have low-risk, early-stage cancer.

Using statistical modelling, the team divided the patientsinto four groups according to patterns of gene activity and,after almost two-and-a-half years, they found patients in onegroup had died significantly earlier than those in the others.

They pinpointed nine key active genes shared by all patientsin that group, and when they tested another 70 Americans withprostate cancer, they again found these genes identifiedpatients who survived for a shorter time - around 9 monthscompared to over 21 months for those without the gene pattern.

The second study by researchers in the United Statesidentified a set of six genes linked to a more aggressive formof prostate cancer in a group of 62 patients at the Dana-FarberCancer Institute in Boston. The signature divided patients intotwo groups: one with an average survival time of 7.8 months andthe other with an average survival of at least 34.9 months.

The British team said their signature included several genesinvolved in the immune system - suggesting the immune system issuppressed in patients whose cancers spread around the body.

Commenting on the work in The Lancet Oncology, KarinaDalsgaard Sorensen at Denmark's Aarhus University Hospital, whowas not involved in either study, said the findings were welcomeand significant.

"These results suggest that a few selected genes in bloodsamples from patients...can significantly improve the predictionof outcomes," she said.

(Reporting by Kate Kelland, editing by Andrew Heavens)

((kate.kelland@thomsonreuters.com)(+44)(0)(207 5420823)(Reuters Messaging: ))