Manufacturing challenge leads to death in CAR-T trial

An unusual case of resistance to CAR-T highlights a challenge researchers face when trying to program a person’s immune system to fight cancer: What to do if a cancerous cell gets programmed too.

In one of the first CAR-T clinical trials, a young man with a hard-to-treat leukemia relapsed and died several months after undergoing CAR-T—a treatment in which his own T cells were engineered to carry a chimeric antigen receptor (CAR) recognizing an antigen, called CD19, on leukemic cells.

In an investigation of his relapse, described in Nature Medicine, a team at the University of Pennsylvania led by physician Marco Ruella found that a single leukemia cell survived the process, undergoing the same CAR engineering as the man’s T cells. While his engineered T cells went back into his body to find and destroy leukemia cells dotted with CD19, the contaminating cancer cell was able to survive by covering up its CD19 with the engineered receptor it wasn’t supposed to have, Ruella says. That cell seeded a proliferation of leukemic cells that eventually killed the patient (Nat. Med.2018, DOI: 10.1038/s41591-018-0201-9).

“If you have the CAR and you have the target”—the CD19—“in the same place, they bind each other,” Ruella says. “The CAR-T cells cannot see that CD19. This is an exceptionally rare event.”

Hundreds of people, mainly with blood cancer, have been treated by CAR-T, either through a clinical trial or through one of two FDA-approved products. Resistance can develop, but typically it is due to mutations in the gene for CD19, Ruella says. In another paper in Nature Medicine, a team from Novartis describes various truncated and mutated CD19 variants that have arisen in patients who develop resistance to CAR-T (Nat. Med.2018, DOI:10.1038/s41591-018-0146-z).

The general CAR-T protocol involves removing the fraction of a patient’s blood that contains immune cells. Researchers then pull out the T cells, infect them with a virus that carries the gene for the CAR, find and grow the ones that successfully make the receptor, and return them to the patient, where they home in on their target. Typically, Ruella says, cells that aren’t T cells die off, including any leukemic cells that slip through.

A spokesperson for Novartis, whose FDA-approved CAR-T treatment Kymriah is derived from the University of Pennsylvania work, says the firm’s manufacturing process differs from the university’s, including steps to eliminate leukemic cells. Another FDA-approved treatment, Yescarta, is sold by Gilead Sciences. The company did not immediately respond to a request for comment.

Since the discovery of the errant leukemic cells, Ruella says, the university has discussed changing its manufacturing process, including examining leukemic cells for the presence of the CAR, better enrichment of T cells, and depletion of the tumor cells. But changes aren’t made lightly.

“You try to keep the manufacturing process as easy as possible, because if you add more steps, there are higher chances that your manufacturing fails, and you cannot deliver the treatment to the patients,” Ruella says. “Cell manufacturing is one of the most controlled and standardized processes in medicine.”