"The wonderful thing about a comprehensive cancer center is that there are people from many different disciplines, pursuing different ideas," says pathologist Darryl Shibata of USC/Norris Cancer Center. Such cross-pollination at Norris has helped Shibata find a new way to learn about how tumors grow.

Shibata peers backwards in time by using information stored in individual tumor cells.

In the same way anthropologists use genetics to trace the evolutionary relationships between different groups of humans, he uses genes to reconstruct the history of particular tumors.

Most tumors become evident to physicians after surreptitiously existing for some time. When pathologists finally examine the tumor cells, they find little information about the tumor's history. But just as paleontologists can learn about dinosaur behavior by looking at fossils, Shibata and his colleagues learn about a tumor's growth and development by examining the genetic changes that have accumulated along the way.

"It's like the 'out of Africa' research that evolutionary biologists are doing," he says. "The entire tumors are the continents and the individual tumor cells are the people living there." Phylogenetics, or the method of looking at genes to determine the history of populations, is a "new tool for cancer," he adds.

"We really don't know much about tumor behavior," says Shibata. But tumor cells, in a sense, "know" which cells they've descended from-a version of cellular memory.

"We take a tumor, and examine it to look at its history," he says. The best way to do this is to look at the small, heritable changes that occur in cancer cell genes as a tumor grows.

The changes, or mutations, serve as markers, just as blood type or eye color serve to identify different groups of people.

Knowing how often certain mutations tend to occur can lead to the development of a "molecular clock," helpful in determining the time frame of a tumor's growth.

The mutations in the genes are passed along as tumor cells divide and multiply, so entire sets of cells with similar mutations arise from the same "ancestor."

For instance, all of the cells from the left half of a tumor might display one type of mutation while cells from the right half reveal another. This information would suggest that early in the tumor's history, two cells, one with each type of mutation, gave rise to two populations of tumor cells-one which formed the left half of the tumor and the other which formed the right. This case would be different from one in which all the cells on the surface of a tumor carried one mutation, while cells deep within carried another.

Tumor phylogenetics could soon help doctors uncover the mysterious past of particular cancers. "If you want to prevent cancer," Shibata says, "you have to know how it starts."