Scientists have mapped the cascade of genetic changes that turn normal cells in the brain and pancreas into two of the most lethal cancers. The result points to a new approach for fighting tumours and maybe even catching them sooner.

Genes blamed for one person's brain tumour were different from the culprits for the next patient, making the puzzle of cancer genetics even more complicated.

But the new research also found that clusters of seemingly disparate genes all work along the same pathways. So instead of today's hunt for drugs that target a single gene, the idea is to target entire pathways that most patients share.

Think of delivering the mail to a single box at the end of the cul-de-sac instead of at every doorstep.

The three studies, published in the journals Science and Nature, are a milestone in cancer genetics.

"Finding those common elements within the landscape suggests there are therapeutic interventions that can help the whole group," said Dr. Phillip Febbo of Duke University's Institute for Genome Sciences and Policy.

Despite 30 years of laborious work, scientists until now have found only a fraction of the genetic alterations required to cause any of the 200 diseases that collectively are called cancer. Different tumours require a different domino effect of genetic changes to arise, and to determine their severity and even which treatments will work.

The new maps do not include just mutated genes. They cite missing ones, extra ones, and overactive or underactive ones, too, in the most comprehensive look ever at human tumours.

Teams led by Johns Hopkins University examined more than 20,000 genes in tumours taken from 24 pancreatic cancer patients and 22 patients with the most dangerous brain tumour, called glioblastoma multiforme. Separately, The Cancer Genome Atlas project -- a government-funded network of 18 medical centres -- analysed 600 genes in glioblastomas from 206 patients.

The Hopkins teams found hundreds of genetic changes, including a particularly intriguing gene named IDH1. Twelve per cent of glioblastoma patients, mostly young ones, harboured a mutated version that meant longer survival .

If more study proves that effect, doctors soon might use an IDH1 test to help decide prognosis, said Hopkins' Dr. Victor Velculescu.

The bigger discovery involved cancer's genetic chaos. No tumours were identical. The typical pancreatic cancer contained 63 genetic alterations and the average brain tumour 60, Hopkins researchers reported in Science.

Fortunately, "genes don't work alone," said Hopkins' Dr. Kenneth Kinzler, who led the pancreatic work. Figure out which genes cluster in which pathways and "a simpler picture emerges."

The Hopkins team identified 12 core pathways that were abnormal in most pancreatic tumours. In Nature, The Cancer Genome Atlas researchers reported three core pathways in most glioblastomas.

Some pathways allow damaged DNA to escape repair. Some switched off protective factors meant to suppress tumours.

Companies already are researching drugs to block an enzyme pathway implicated in the studies.

"Even though it sounds complex, it's actually allowing us to simplify the complex into pathways that will allow us, I think, to truly understand cancer for the first time and take a much more rational approach to treatment," said Dr. Anna Barker of the National Cancer Institute, who co-directs the cancer atlas project.

Moreover, the work suggests possible ways to catch cancer earlier, by tracing mutant DNA floating through the bloodstream well before tumours themselves start to spread.

As one expert said: "I don't think that's any longer science fiction."