A newly discovered group of cells present in colorectal tumors may be the source of the cancer's ability to grow unchecked. Researchers found that tumor cells containing a combination of three proteins were capable of becoming tumors again when injected into mice. They speculate that the cells may be offshoots of healthy stem cells that no longer respond to the body's natural checks against aggressive, uncontrolled growth.

Researchers believe that tumors grow from clumps of cells that mimic stem cells' ability to reproduce indefinitely and grow tissue made of several kinds of cells—possibly because the mimics were once healthy stem cells, according to the "cancer stem cell" hypothesis. Stem cells or not, these cancer linchpins should grow tumors when transplanted into other animals in the same way that normal stem cells can regrow healthy tissue.

Zapping them directly may be the key to stopping cancer in its tracks. "If you want to eradicate a tumor in a patient, probably you have to target the [cell] population that is responsible for the tumor growth," says oncologist Piero Dalerba of the Stanford Institute for Stem Cell Biology and Regenerative Medicine.

Stemlike cancer cells have already turned up in solid tumors of the breast, pancreas and head or neck. And last year, two research groups reported similar cells in colon tumors that produced a marker protein on their surfaces called CD133. Not all colon cancer cells exhibit CD133, however, so Dalerba and colleagues sought out more specific markers from six colon tumors surgically removed from people or mice and scattered into individual cells. They then picked out the proteins EpCAM and CD44, which help cells stick to each other.

The team injected cells showing the two markers under the skin of mice and found that they grew into large tumors similar to the original tumors up to 80 percent of the time [see image above], according to a paper published online today by Proceedings of the National Academy of Sciences USA.

Next they looked at cells that also had a third marker, CD166, which Dalerba says was particularly intriguing because patients whose tumors have high concentrations of this protein are less likely to survive. Cells with either EpCAM or CD44 that lacked CD166 did not grow tumors, they found, whereas cells containing all three markers grew a tumor even though injected in low numbers.

"They've identified CD166 as a potentially interesting marker," says cancer biologist Richard Hill of Princess Margaret Hospital in Toronto, but "I don't think the data is terribly convincing yet. … They really only looked at it in two or three tumors." Moreover, he adds, the study does not yet prove that the marked cells initiate or sustain cancer.

The function of the markers is unclear, agrees Dalerba, who compares them with spots on a dog. "They don't tell us specific information about the nature of the dog," he says. "They allow us to recognize special breeds" and open the door to learning what makes them bark.