As an idea for wiping out cancer, it could have been ripped from the pages of a spy thriller: Take cancer cells that have departed the original tumor and spread elsewhere in the body, genome-edit them to be stone-cold killers, then wait for the homesick cells to return and make like émigré assassins.

In a study four years in the making, scientists reported on Wednesday that “rehoming” cells that had been CRISPR’d to attack cells in the original tumor improved survival in lab mice with brain cancer, as well as in mice with breast cancer that spread to the brain.

That cancer cells migrate back to the original tumor after metastasizing to distant sites is still, 12 years after its discovery, one of the most unexpected and perplexing in cancer biology. Called self-seeding or (redundantly) rehoming, the surprising behavior has inspired several treatment ideas, such as putting cancer-cell-killing viruses or suicide genes into the rehoming cells (which would somehow have to be made resistant to those lethal agents) and hoping they transfer their lethal payload to the tumor cells they find when they return home.

Most of those approaches have stumbled, and the new study, published in Science Translational Medicine, is very preliminary and comes with the usual caveats. If experimental cancer therapies worked as well in people as they do in lab mice, the War on Cancer would have been won years ago. Instead, the chance that an experimental cancer therapy will become an approved therapy is about 5 percent. And that’s of the ones that enter clinical trials. Many treatment ideas fail even earlier.

Biologist Khalid Shah, director of the Center for Stem Cell Therapeutics and Imaging at Brigham and Women’s Hospital and the leader of the study, said he’s not deterred by the challenge. Since rehoming tumor cells “can track the original tumors, it is a matter of taming these cells to find the ultimate cure,” he said.

He’s planning a startup company to turn rehoming cells into cancer killers, Shah said, adding that in the year since the new study was submitted to the journal, he and his colleagues have obtained more data on how well CRISPR’d rehoming tumor cells can attack cancer: “We’re in it for the long term.”

Other scientists who have investigated rehoming cancer cells said they can imagine a therapy based on these returning émigrés looking promising enough to enter clinical trials. “With sufficient resources and sophisticated expertise, one could see this being developed toward a Phase 1 trial,” said cancer biologist Renata Pasqualini of Rutgers University. She led a 2016 study that used rehoming cancer cells to deliver an anti-tumor compound directly to primary and metastatic tumors in mice; that reduced the tumors’ growth, damaged the blood vessels they need to survive, and triggered a self-destruct mechanism in many of the cells. The chance of her approach or Shah’s being tested in people, she cautioned, “really depends on safety being demonstrated.”

For their new study, Shah and his colleagues first removed tumor cells from mice. (If the approach ever makes it to patients, isolating cells from a biopsy or from a surgically removed tumor would also be the first step.) They then used the genome-editing technology CRISPR to make the tumor cells express a molecule that activates the aptly named “death receptor” on cancer cells, causing the cells to self-destruct. Any genome-editing technology would probably work, Shah said, but CRISPR is the easiest.

Injected into mice, the cells made their way to all three types of tumors the scientists tested: primary glioblastoma, the most lethal form of brain cancer; recurrent glioblastoma, in which the cancer was treated but returned because it became resistant to the standard chemotherapy; and breast cancer that had metastasized to the brain. The rehoming cells brought “marked survival benefits,” the scientists wrote.

In mice with primary or recurring brain cancer, tumors shrank considerably and 90 percent of the rodents survived for weeks or monthsafter treatment, Shah said; usually such glioblastomas are nearly 100 percent fatal in both mice and people, and were so in untreated mice in the experiment. In mice with the brain metastases, about half survived for several weeks after the cell therapy — evidence that engineered rehoming tumor cells might be therapeutic.

A key unanswered question is what happens to distant metastases even if CRISPR’d rehoming cells successfully attack the original tumor. As far as biologists know, they make a one-way trip, to the primary tumor but not to distant metastases, which can occur in multiple organs. Those metastases, not the original tumor, are responsible for upwards of 90 percent of cancer deaths. Another question is how to keep CRISPR’d rehoming cells from going rogue and initiating new tumors rather than killing the original one; CRISPR’d or not, they’re still cancer cells.

Shah and his team gave the engineered cells a kill switch so that administering a simple drug would kill them, presumably after they had eliminated cancer cells and before they had seeded any new tumors. It remains to be seen how well that would work in people.

Republished with permission from STAT. This article originally appeared on July 11, 2018