The blood-brain barrier disruption appears to be “a unique mechanism within this setting,” said Mark Gilbert, Juno’s chief medical officer, who presented the company’s report.

Juno believes that multiple variables, taken together, led to the devastating effects on patients’ brains, though the company hasn’t elucidated them all in detail. Its investigation focused on the medical history of the patients who died, how their T cells were handled, and what happened to those cells after they were reinfused in the body.

Patients who succumbed to cerebral edema tended to be younger than 30 years old and had intensive chemotherapy with fludarabine and cyclophosphamide before CAR-T cell infusion, Juno’s Gilbert says. But another important factor was probably in the cells:  CAR-T cell therapy is highly personalized, with a new batch made for every patient. And Gilbert acknowledges that variability of the patient’s immune cells before being engineered and perhaps inconsistencies in how the cells were handled throughout the process may have played a role. For example, each patient batch varied in the amount of CD8 cells, a T-cell subtype, and the dose of this kind of CAR-T cell significantly correlated with edema development.

Then there’s what happens after the T cells are given back. The Juno investigation found that when the CAR-T cells multiplied extremely rapidly, the patient was more likely to suffer severe neurotoxicity and edema. The modified immune cells typically proliferate about 12 to 14 days after a patient receives them. But, in the fatal cases of the ROCKET trial, this expansion happened a week sooner, only 6 to 8 days postinfusion. Individual variability in blood levels of IL-15, a molecule that stimulates T-cell growth, may account for the early CAR-T surge; IL-15 levels in the patients who had the surge were both higher at the start and grew faster than those in other patients receiving CAR-T cells. Previous research shows IL-15 levels correlate with antileukemia activity, but also with neurotoxicity.

Together, the various patient and product factors “create cumulative risk for fatal neurotoxicity,” says Gilbert, although he cautions that the results of Juno’s internal investigation are not definitive.

Yet, it is already using lessons from the ROCKET trial in their latest product, a CAR-T therapy for adults with relapsed or refractory non-Hodgkin lymphoma. In its current phase I clinical trials, the T-cell subtypes are monitored.

Sadly, fatalities are always a risk when developing therapies for advanced cancers, Gottschalk says. “This expectation that you can develop therapies for life-threatening diseases [without deaths] is rather naïve,” he says. “Hopefully, [Juno’s investigation] will also inform other studies going forward.”  When asked whether the deaths in the ROCKET trial could happen with other CAR-T therapies, he doesn’t mince words. “In the end, yes.”