Scientists at the Stanford University School of Medicine have identified a cancer-initiating cell in human melanomas. The finding is significant because the existence of such a cell in the aggressive skin cancer has been a source of debate. It also might explain why current immunotherapies are largely unsuccessful in preventing recurrence of the disease.

"These cells lack the traditional melanoma cell surface markers targeted by these treatments," says post-doctoral fellow Alexander Boiko. "Without wiping out the cells at the root of the cancer, the treatment will fail."

Boiko is the first author of the research, which was published in the July 1 issue of Nature. He works in the laboratory of Dr. Irving Weissman, director of Stanford's Institute for Stem Cell Biology and Regenerative Medicine. Weissman is the senior author of the research.

The cancer stem cell theory holds that, like queen bees in a hive, only a subset of cancer cells are at the root of the tumor's growth. These cells can both self-renew (that is, make more of themselves) and differentiate into other tumor cell types.

Any therapy that doesn't wipe out the cancer stem, or initiating, cells has no chance of completely eradicating the disease even if it destroys nearly all other tumor cells. That's why, say proponents, it can be relatively easy to get a patient into remission, but extremely difficult to prevent the cancer stem cells from roaring back and causing a relapse months or years later.

Cancer stem cells first were identified in blood cancers, but since then have been identified in a number of solid tumors, including in bladder, brain, breast, and colon cancers. Previous studies in the laboratory of assistant professor of radiation oncology Dr. Maximilian Diehn, in collaboration with the labs of Weissman and Stanford colleague Dr. Michael Clarke, have indicated that cancer stem cells may be more resistant than other cancer cells to many common treatments like radiation and some chemotherapies.

Although a growing body of evidence seems to support the cancer stem cell hypothesis, melanoma has remained a conundrum. A University of Michigan study in 2008 found that as many as one in four melanoma cells could cause cancers in immune compromised mice, suggesting there might not be a particularly privileged subset of cancer stem cells in this tumor type. Boiko set out to solve the mystery.

"I didn't know if melanoma would in fact have the cancer-initiating cells," says Boiko. "I was completely unbiased, so I was actually sort of surprised to find such a clear-cut answer. It fits exactly what's been discovered in the studies of other solid tumors."

To conduct the study, Boiko analyzed cell surface markers on primary melanoma tumor samples taken directly from patients at the Stanford Cancer Center. In this way, he avoided having to grow the cells for a long period of time in the lab. Continuous culturing of cancer cells often gives the cells time to evolve and change in ways that might not accurately reflect their composition in melanoma patients.

He found that one protein, CD271, always was expressed on only a fraction of the cells in the human melanoma samples tested: The proportion of cells expressing CD271 varied in the samples from 2.5 to 41 percent of the total cell population; the marker appeared on a mean of 16.7 percent of cells in the samples.

That was interesting, because CD271 previously was found to be a marker that identifies a group of cells called the neural crest stem cells. These cells are unique in that they are a migratory cell population that becomes many cell types during development, including melanocytes (cells responsible for skin pigmentation), bone, smooth muscle, neurons, and cartilage in the head and face.

When Boiko transplanted the melanoma cells from nine human samples into laboratory mice with severely compromised immune systems, he found that the cells expressing CD271 on their surface were much more likely to cause cancers in the recipients than those from the same tumor that didn't express the marker (70 percent versus 7 percent, respectively). All but one of the newly induced tumors arising from the transplantation of the CD271-positive cells went on to develop a population of a mixture of CD271-expressing and non-expressing cells—indicating that the cells with the marker were both self-renewing and differentiating into other types of tumor cells.

Boiko then collaborated with researchers in the medical school laboratories of professor of surgery Dr. Michael Longaker and assistant professor of surgery Dr. George Yang to test further the tumor initiating properties of the cells that expressed CD271. They transplanted normal human skin onto the backs of the immunocompromised mice and injected the skin with the melanoma cells. Only cells expressing CD271 (isolated from melanomas from two patients) gave rise to tumors and lung metastasis in the mice.

Finally, the researchers looked to see whether the cancer-initiating cells also expressed common cellular antigens currently used for melanoma therapy. They found that melanoma cells expressing CD271 either completely or partially lacked expression of three common therapeutic targets — TYR, MART, and MAGE — in 86 percent, 69 percent, and 68 percent of melanoma patients, respectively.

"This could be the reason why we often see melanoma patients relapsing and coming back to the clinic," says Boiko. "Our research indicates that it may be more appropriate to also target cells expressing CD271." Such a combination therapy might work to kill both types of cells in the tumor and, hopefully, prevent disease recurrence.

So why do some melanoma tumors seem to have such a large proportion of cancer-initiating cells?Boiko and his colleagues speculate that the answer might lie in the rapidly evolving, aggressive nature of the disease. It's possible that a kind of natural selection among the dividing cancer cells occurs that, in extreme cases selects for one clone, or cell lineage, that for the most part fails to differentiate into non-tumor-initiating cells.

"Extreme care should be taken to identify the origin and the past laboratory history of the cells under study," says Boiko. "Our work was done on primary and minimally passaged patient samples. The variations that we saw in terms of CD271 prevalence in these samples could depend on the stage of the disease the patient was in, and the aggressiveness of that individual cancer."

The research was supported partly by the National Institutes of Health, the American Cancer Society, and the Virginia & D.K. Ludwig Fund for Cancer Research.