Cancer complexity slows quest for cure

Genomic analysis reveals multiple mutations in tumours.

Solid tumours such as glioblastoma (red) can be caused by multiple genes in different patients. Credit: CNRI/SPL

Hopes that large studies of cancer genomics will justify their high cost by offering a fast-track to cures have been dealt a blow by a series of papers.

The controversial Cancer Genome Atlas, run by the US National Institutes of Health, is analysing genetic and epigenetic changes in cancers. Still in its pilot phase, the project could eventually cost up to $1.35 billion. But it does not include 'functional' studies that investigate how the mutations aid tumour development and what drugs might target the pathways essential to tumour survival.

Yet functional studies are exactly what two papers, from a group led by Bert Vogelstein of the Johns Hopkins School of Medicine in Baltimore, Maryland, suggest are needed. He and others say that the focus should shift from hunting for individual genes that cause certain cancers, to disrupting the broader biological pathways that support cancer growth.

"It is apparent from studies like ours that it is going to be even more difficult than expected to derive real cures," says Vogelstein.

Two papers from his team, plus one from the Cancer Genome Atlas group, appeared on 4 September in Science¹_,² and Nature³. The studies find that individual cancer patients each carry dozens of genetic mutations — an average of 63 alterations in pancreatic cancer and 47 DNA mutations in one type of brain cancer. Similar results have been found in previous studies of other cancers. This makes it unlikely that the cancers will be cured by drugs that target just one or a few genes, the researchers say.

Other scientists call the latest results sobering but important. "Unfortunately this wasn't what we all hoped for," says Stephen Elledge of Brigham and Women's Hospital in Boston, Massachusetts. "But there is useful information in there, and what they're learning is more of what they already learned — which is that cancers are extremely complex."

Vogelstein's team analysed nearly 21,000 genes in some 20 patients with a type of brain cancer called glioblastoma multiforme¹ and a similar number with pancreatic cancer². The Cancer Genome Atlas team looked at 600 genes in 91 patients with glio­blastoma multiforme³.

What they’re learning is more of what they already learned — which is that cancers are extremely complex.

The studies hoped to find genetic glitches driving the tumours that could be treated by drugs similar to Novartis's Gleevec (imatinib) and Genentech's Tarceva (erlotinib), which inhibit the activity of some mutated genes that can cause certain cancers. Instead, the findings confirm earlier hints that patients with the same cancer diagnosis can harbour different sets of genetic causes.

"It is extremely unlikely that drugs that target a single gene, such as Gleevec, will be active against a major fraction of solid tumours," says Vogelstein, whose group has published similar studies on breast and colorectal cancers⁴.

The latest papers do identify some single genes that seem to be important in subsets of the cancers. For instance, Vogelstein and his colleagues report that a gene called IDH1, which had not previously been linked to brain cancer, was often mutated in younger patients with a certain type of glioblastoma. And the Cancer Genome Atlas group reports that a gene called NF1, whose link to cancer had been hypothesized before, was mutated in 23% of 206 patients that were analysed.

The atlas group also found that patients with a particular epigenetic make-up who are treated with one type of chemotherapy show a pattern of genetic and epigenetic changes that may render them resistant to further treatments. The mutation pattern suggests why this resistance evolves and may help doctors find strategies to avoid it.

That shows the importance of the atlas and other similar studies, such as those included in the 10-nation International Cancer Genome Consortium, says Lynda Chin of the Dana-Farber Cancer Institute in Boston, who was the team leader for the atlas paper. "These are very important, clinically relevant questions, and you can't answer them in a traditional hypothesis-driven manner," Chin says.

The Cancer Genome Atlas has moved more slowly than its architects had hoped because it has proved harder than expected to find enough high-quality tissue samples to analyse. The Nature paper is an interim analysis of the samples it has studied so far. But most of the genes identified in it and the Vogelstein studies had already been identified in earlier studies, and are not found in most patients with a particular tumour.

That bolsters critics of the atlas, such as Elledge, who has long said that functional studies will be needed to filter the most clinically relevant drug targets out of the massive pool of mutations found in cancer.

"The information they're getting is useful," he says, "but it's expensive and I think some of that money should go to help get you further along into finding drugs."