Genome analysis used to decode brain cancer: study

US scientists have unveiled the most complete genetic profile ever attempted of glioblastoma, a common and deadly form of the brain cancer that US Senator Edward Kennedy is battling.

The research uncovered a host of genetic alterations linked to the disease, including three previously unknown mutations found in at least three-quarters of tumour samples analysed.

It may also explain why, in some cases, standard chemotherapy treatment for brain tumours works at first but then loses its ability to beat back tumour growth.

The study, published in the British journal Nature, underscores a powerful new way to study cancer: harnessing powerful computers to analyse not just suspect genes in tissue samples, but entire genomes -- each with tens of thousands of genes -- across hundreds of samples.

"The ultimate aim is to achieve a complete atlas for the genomic changes in glioblastoma," explained Lynda Chin, a researcher at the Dana-Farber Cancer Institute in Boston and a member of the Cancer Genome Atlas Research Network (TCGA), which collectively authored the study.

"It would be essentially like having a full parts list for cars so that you know everything that could go wrong," she said in an interview.

The most aggressive and common of all brain tumours, glioblastoma strikes more than 20,000 people every year in the United States, and accounts for some 13,000 deaths.

Senator Kennedy, who spoke at the Democratic Party convention last month, was diagnosed with the disease this year.

Working with brain tumour samples donated by 206 patients from across the United States, the TCGA group sequenced 601 suspect genes and compared them to the same genes from healthy tissue.

Three new genes not previously linked to glioblastoma were found guilty by association: NF1, also implicated in an inherited disorder that causes runaway tissue growth along nerves; ERBB2, closely associated with breast cancer; and PI3, known to play a role in several cancers.

Besides checking for small mutations in genes, they looked for anomolies that can promote tumour growth: large strings of missing or duplicated genetic code, problems in the way information in the DNA is "transcribed" to produce proteins, and how certain molecules -- called methyl groups -- interact with DNA.

They also evaluated the impact of treatments, to see what kind of changes -- intended or not -- certain drugs trigger.

The most unexpected finding, said Chin, was that a commonly used chemotherapy treatment called temozolomide may provoke a "resistance mechanism" that compromises genes critical for DNA repair.

The result is new tumours with a large number of DNA mutations and a higher tolerance for chemotherapy.

"This type of comprehensive, coordinated analysis of unprecedented multi-dimensional data is made possible by advanced technologies" that simply did not exist a decade ago, said John Niederhuber, Director of the National Cancer Institute.

The TCGA study, which analysed many samples but relatively few genes, was released at the same time as another study, published in the US journal Science, that coded all the genes in 22 brain tumours.

"You can make important discoveries with either approach, but ideally you want to take the strength of both," Chin told AFP.

A new generation of technologies known as single molecule sequencing is "an order of magnitude cheaper and faster" and should make that possible within several years, she said.

"We have the tumours in the bank -- when this technology comes on line, we can go back and sequence not just 600 genes but all of them, and not in 20 samples but 200 or, better yet, 500 or 1000."

The examples highlighted in what she called an interim study are "just the tip of the iceberg," she added.

The TCGA consortium has already initiated the same sort of comprehensive approach to ovarian and lung cancer.