Blood reveals signs of pancreatic cancer.

By the time you find out, it's usually too late. Almost all people diagnosed with pancreatic cancer succumb quickly to the disease, which spreads aggressively to the liver and other organs.

Researchers have for the first time created mice that appear to develop the disease in the same way that people do. The animals offer a "very faithful model of pancreatic cancer," says David Tuveson of the University of Pennsylvania in Philadelphia, who led the new work. The genetically engineered rodents should aid the identification and testing of therapies for the deadly cancer, say Tuveson and his colleagues.

The scientists have already laid the groundwork for a test that could catch the disease in time for potential treatments to be effective. They've found a pattern of proteins in the blood of mice with precancerous pancreatic lesions.

How pancreatic cancer arises has remained obscure because it's typically detected at an advanced stage. "We have never known what the precursor lesion is in people," says Tuveson.

He and his colleagues created mice in which the cells that give rise to the pancreas contain the most common mutation found in human pancreatic cancer. The mutation occurs in a gene called KRAS, which is defective in more than 90 percent of people with pancreatic cancer.

Within months of birth, the mutant rodents developed lesions in pancreatic ductal cells that are similar to lesions that physicians have seen in the pancreases of people with and without cancer, Tuveson and his colleagues report in an upcoming Cancer Cell. Scientists have speculated that these lesions are comparable to the polyps that presage colon cancer. Indeed, more-severe lesions appeared as the mice aged, and, in a few old mice, the lesions gave rise to highly malignant tumors.

A team headed by Ronald DePinho of Dana-Farber Cancer Institute in Boston has crossed the KRAS-mutant mice with mice lacking a tumor-suppressor gene implicated in pancreatic cancer. The offspring developed invasive pancreatic tumors even more quickly than Tuveson's mice did, DePinho's team reports in the Dec. 15 Genes & Development.

By studying the pancreatic lesions in mice, Tuveson and his colleagues have identified cellular molecules that could be targeted by anticancer drugs. They're also developing tests to reveal the lesions. "What we're hoping is that the mouse will be a springboard to new ways of finding these lesions in people," says Tuveson.

Investigators at the Food and Drug Administration-National Cancer Institute Clinical Proteomics Program in Bethesda, Md., have already analyzed blood samples from the KRAS-mutant mice. They found a pattern of protein abundances that clearly distinguishes healthy mice from those with pancreatic lesions. "What's fascinating now is that we actually have a premalignant pancreatic cancer signature," says Emanuel Petricoin of the FDA, codirector of the program.

Investigators have begun looking for this signature in people diagnosed with pancreatic cancer. "The results are promising," says Lance Liotta of the NCI, the program's other codirector.

Liotta and Petricoin contend that people at high risk of the disease could benefit from such a diagnostic tool. Pancreatic cancer is so rare that it's unlikely that a screening test for the general population would be accurate enough to be useful, they say.