Why this could be strike three for Juno’s lead

Two clinical holds after multiple deaths with Juno’s lead CAR-T asset, JCAR015, might have been containable, but a third – revealed today – could be its end.

In hindsight the voluntary hold fundamentally calls into question the FDA’s readiness to resume JCAR015 studies in July after just five days. While a fresh delay is now inevitable Juno does have a different anti-CD19 asset to pivot to, but with its market cap losing $1bn this morning the problem could be toxic for Juno and its competitors alike.

For instance, while Kite’s KTE-C19 is a different construct and has not been tainted by toxicity to this extent, it will soon go before the FDA in a regulatory filing. If nothing else, the agency will surely now take a much tougher stance given how much is still not known about the risks of CAR-T; Kite opened down 5% today before rising.

Not discontinued... yet

For now Juno says it is not discontinuing JCAR015, a project derived from its collaboration with Memorial Sloan-Kettering Cancer Center, but on a call today its chief executive, Hans Bishop, accepted that this was an option; an update will be given at an investor event at next month’s Ash meeting.

It was remarkable how little light the company shed on this latest setback, which involved the deaths this week of two patients from cerebral oedema in the pivotal Rocket study in adult ALL. Mr Bishop said 12 patients had been dosed since Rocket resumed in August, and the two who died most recently had been dosed last week.

When Rocket was halted in July after two other deaths the blame was put on a fludarabine lymphodepleting regimen, and the trial restarted with cyclophosphamide-only conditioning. The theory that fludarabine was to blame can thus be dismissed, and all Mr Bishop could offer by way of explanation was that the issue was “multifactorial”, and could be down to the construct or the disease.

If the issue did not threaten to engulf all CAR-T players it could be argued that those who continued to use fludarabine-based chemo – there are many – were right to do so, and no longer have this specific issue to worry about. Juno said the deaths did not correlate with the manufacturing site (Rocket uses two) or process.

Remarkably, it insisted that restarting Rocket quickly without fludarabine in August was the right decision. No doubt personal injury lawyers are paying close attention.

Quick pivot?

At least Juno does have the option of pivoting to a backup project, JCAR017 derived from Seattle Children’s Hospital, whose lead indications are paediatric ALL and lymphoma. For some time now the group has been stressing JCAR017’s potential.

There are important differences: while JCAR015 uses a CD28 co-stimulatory domain and gamma-retroviral transfection, JCAR017 is based on 4-1BB and a lentiviral system, as well as using a defined composition of T cells. The kinetics of cell expansion using 4-1BB are slower, Juno said, and a much more uniform and predictable product results from a defined-composition approach.

Juno was also eager to highlight that it had seen no treatment-related deaths in JCAR017 trials, but cautioned that this construct had not been tested in adult ALL patients. Assuming no delay to Kite or Novartis, switching to JCAR017 would put Juno at least a year behind its competitors, with a filing not coming until 2018.

There could be hidden benefits: an inter-partes review petition filed by Kite challenges the patent Juno licensed from Memorial on which JCAR015 relies, whereas JCAR017 is relatively unencumbered. It also seems that future earnouts on JCAR017 are capped at $95.5m to Seattle Children’s, against $150m to Memorial on JCAR015.

Of course, considering patient deaths and Juno’s inability to reassure the markets that the problem is containable, these are minor considerations. A formal FDA hold on JCAR015 is inevitable, which would make it the third such halt since 2014 (Neurotoxicity only short-term toxic to Juno stock, July 13, 2016).

The first was lifted after about a month, the second after five days. It might take a lot longer to draw a line under the third.