Juno uses lethal neurotox lessons to guide Goldilocks formula for its next-gen CAR-T
WASHINGTON, DC — Neurotoxicity killed JCAR015 right along with 5 patients in the lead pivotal study, derailing Juno Therapeutics $JUNO and allowing first Novartis and then Gilead/Kite to surge way ahead with the first two approvals in the field. But Juno now plans to take what it’s learned from that lethal implosion and make over its next drug — JCAR017 — into what it promises will be a dramatically improved drug that can leapfrog the leaders.

Mark Gilbert, the chief medical officer at Juno, appeared at the Society for Immunotherapy of Cancer (#SITC2017) meeting over the weekend to outline where the biotech has reached in its investigation, and how it’s applying those lessons to make JCAR017 into the impressive therapy that they’ve begun updating along the way to a planned marketing application.
“We always thought it was multiple things that were contributing,” Gilbert told me in a one-on-one at SITC. “We just couldn’t identify them as well as we can today.”
That failure led to one of the most deadly reactions tracked in the clinic in recent biotech times.
Initially, Juno halted its study of JCAR015 after a pair of deaths from brain swelling — cerebral edema — triggered a safety alert, but almost immediately got a green light from the FDA to get back into the clinic after pinning the blame on fludarabine, which is used to prep patients for the personalized cell therapy.
Once dosing began again, though, another three patients died in quick order, killed by the same neurotoxic reaction to the therapy and forcing Juno back to the drawing board to understand better what went wrong, and how to control it in their next ongoing program for JCAR017.
“The early and rapid expansion (of cells) we saw appears to correlate with higher levels of IL-15,” a growth factor for T cells, says Gilbert. “We now know that IL-15 was increased before they ever got CAR-T cells.” And that was associated with the combined use of fludaribine and cyclophosphamide.
“We like to see CAR-T cell expansion that peaks out 11, 12, 14 days,” he says, looking for a steady Goldilocks approach to the ramp up — neither too hot or too cold. “We see that in JCAR017 in TRANSCEND. In ROCKET that happens in 7 days in some of these patients. That expansion is what we want to avoid.”
“As paradoxical as it sounds,” he adds, “the issue wasn’t so much fludaribine as the two drugs together. We’ve moved to a substantially less intensive lymphodepletion regimen.”
But there’s more. Juno has been working on delivering a more personalized cell therapy, looking to assign special levels of therapy calibrated to a variety of factors, including the type of cancer patients have, their age, the level of disease burden a patient has or the presence of antigens.
“Different disease settings have risk of early expansion,” he adds.
Juno — and some of the other companies in this area, including Novartis — also notes that there is animal data to suggest that the 4-1BB costimulatory domain now in use could also play a role in the pharmacokinetics, which is why moving from CD28 on JCAR015 to 4-1BB in JCAR017 is being watched closely. Ultimately, that could also have a bearing on Gilead’s pioneering CAR-T Yescarta, which uses CD28.
CAR-T has always been a high-risk, high-reward kind of therapy, reserved initially for some very sick patients. Cellectis, with its off-the-shelf approach, has just come out of its own two-month clinical hold. As the therapies become more mainstream, that risk has to be better controlled, with the companies that are best at manufacturing and delivery looking for a dominant market share.
Juno may have been badly delayed. But it’s still in the bigger race now underway.