A new type of cancer therapy—recently approved by the U.S. Food and Drug Administration for children with leukemia who run out of options—is changing medical practice and has triggered excitement in the normally reserved world of cancer science.

The new therapy, called Kymriah, comes from pharmaceutical giant Novartis. It is the first in a novel class of treatments known as CAR-Ts—for chimeric antigen receptor T cells—in which a patient’s own immune cells are withdrawn, engineered to target cancer cells, and then infused back into the body. “CAR-T cells are going to play a major role in the blood cancers, because their benefits hugely outweigh their toxicities,” says Steven Rosenberg, chief of the Surgery Branch at the National Cancer Institute, who administered the first-ever CAR-T treatments beginning in 2009. Eleven of those 12 initial adult patients are still alive, despite having terrible prognoses before the therapy. CAR-Ts “can be curative, it appears,” Rosenberg says.

But some patients still suffer significant side effects. One in a clinical trial for a different company’s experimental CAR-T treatment died late last month, nine days after getting the therapy—raising questions about how much researchers understand about CAR-Ts and how far the therapy still has to improve to become widely used. A half dozen experts in the field interviewed in person, along with others who spoke earlier this month at the CAR-TCR Summit in Boston, say they hope to boost both CAR-Ts’ effectiveness and safety with measures such as lowering doses, adding “control switches” and delivering treatments earlier in the disease process.

Plus Kymriah is more expensive than many of even the costliest cancer therapies, so far priced at $475,000 for the onetime personalized procedure—and that price could climb still higher to treat other cancers, industry leaders said last week. “The real answer [on pricing] will come in the coming two to three years, as we see how many people can enroll into this type of therapy and…pay this price,” André Choulika, CEO of biopharmaceutical company Cellectis, said in a talk at the summit. “Nobody knows how much value people will put on this.”

Kymriah’s federal approval on August 30 means pediatric oncologists have to gear up to treat more people, now that they can offer CAR-T to all their advanced acute lymphocytic leukemia patients—not just those who meet the criteria for clinical trials—says Kevin Curran, a pediatric oncologist at Memorial Sloan Kettering Cancer Center in New York City.

Roughly 600 U.S. children a year have acute lymphocytic leukemia that does not respond to two or more other therapies. So far, Curran says, clinical trials have found that 70 to 90 percent of these children can be successfully treated with CAR-Ts, leaving them with no remaining evidence of cancer after one treatment. These results have created tremendous optimism in the field, Curran says, adding that he anticipates CAR-Ts will soon be used beyond acute lymphocytic leukemia. “We don’t think that’s where it’s going to stop,” he says.

The U.S. Food and Drug Administration is shortly expected to approve Kymriah and a similar therapy from Kite Pharma, currently called axicabtagene ciloleucel, for use in some non-Hodgkin lymphoma patients. But Curran and others were quick to note that CAR-T treatments can still be significantly improved.

Adding some kind of control switch to the engineered cells would allow doctors to dial down the therapy if a patient’s immune system gets overactivated, or to dial it up if the treatment is not having the desired effect, says Sattva Neelapu, a professor and lymphoma expert at The University of Texas M. D. Anderson Cancer Center. This might allow doctors to deliver a safer, lower dose of the cells rather than the large and sudden infusion used now, he says. Such controls have not yet become an approved part of the treatment process, however. Cellectis’s Choulika said at the summit that a switch probably would not have been able to activate quickly enough to save the trial patient who died.

Researchers are also wondering whether some patients might need booster doses of their own modified cells to keep their cancer under control after the initial treatment, or of cells engineered to go after more than one target on a tumor cell.

Roughly 80 percent or more of patients with acute lymphocytic leukemia go into remission with CAR-Ts but 20 to 50 percent ultimately suffer a relapse for reasons that are unclear, says David Porter, a professor of medicine and director of the Blood and Marrow Transplantation program at the University of Pennsylvania. “We need to study not just how to treat relapse but why do patients relapse,” he says. “If you can understand that, you can start to prevent subsequent recurrence. That’s a major goal of future trials in [acute lymphocytic leukemia].”

In lymphoma patients relapse is less likely after CAR-T therapy, Porter notes, but fewer achieve remission. “Why do some patients respond and others do not?” he says. “There needs to be some very fast and important research done to understand the mechanisms of response or resistance, so this therapy can be developed to be more effective for more patients.”

Another major issue, of course, is safety. Although Rosenberg says side effects with CAR-T are often less severe than with chemotherapy, the most common one is cytokine release syndrome (CRS)—an activation of the immune system that feels like a terrible case of the flu and in very rare cases can lead to death. Most patients weather the storm in a hospital’s intensive care unit with a high fever, aches, fatigue and sometimes hallucinations.

Novartis, which developed Kymriah in collaboration with researchers at Penn and The Children's Hospital of Philadelphia, says Kymriah has a “well-characterized and generally manageable safety profile with the application of specific management guidelines and when executed at sites with appropriate training,” according to an e-mailed statement attributed to David Lebwohl, senior vice president and global program head of CAR-T.

But other CAR-T treatments may still pose major risks. The FDA put two small, early-stage Cellectis CAR-T clinical trials on hold earlier this month when the first patient in one of the trials, a 78-year-old man, died after suffering from CRS. In a press release the company said it is considering lowering the amount of engineered cells it infuses in future trial patients. “Cellectis is working closely with the investigators and the FDA in order to resume the trials with an amended protocol including a lowered dosing,” it wrote. Cellectis’s experimental treatment is different from other CAR-Ts—instead of altering patients’ own immune cells to target cancer cells, Cellectis’s CAR-T is “off the shelf,” with all patients getting the same cells that are engineered to attack tumors.

In some patients the cytokine ordeal has been linked to effectiveness; the more miserable they get, the more benefit they seem to receive from the treatment. But Marcela Maus, who directs the Cellular Immunotherapy program at Massachusetts General Hospital in Boston, says patients don’t need to suffer to get long-lasting relief from their cancer.

Doctors hope the cytokine response will be reduced if CAR-Ts are delivered earlier in the disease’s course, before patients have been heavily treated with chemotherapy and when their immune systems are still relatively healthy. Earlier treatment could also save patients months if not years of miserable chemotherapy sessions, which are now the standard approach for patients whose first treatments do not work, Maus says.

Drug companies’ greatest hope is that CAR-Ts will eventually work as well on solid tumors as they do on liquid ones—but that remains a far-off dream for the moment. “Solid tumors by definition are much harder,” says Sunil Agarwal, president of R&D for Juno Therapeutics, another company leading CAR-T development. These tumors have more defenses than blood cancers, making it tougher for CAR-Ts to get through, Agarwal explains. And they lack a distinctive marker that would allow only cancer cells—and not crucial healthy ones—to be targeted and killed. Companies and researchers are now trying CAR-Ts in combination with other treatments such as the other major type of immune therapy, called checkpoint blockades.

Maus says that when she started testing CAR-Ts on glioblastoma, a notoriously lethal form of brain tumor, she was convinced they would quickly provide a cure. Now, after watching years of failures and limited successes, she approaches the task with more humility—but still some optimism. “We could get lucky,” she says.

Porter, at Penn, says he sees the FDA approval as the beginning of a new approach to therapy rather than the end of a long development process. “It now allows you to move faster and study things that need to be done to make this better and safer,” he says. “There is so much to do.”