When five patients with a difficult type of leukemia died last year in a clinical trial run by Juno Therapeutics, it was a big setback for the Seattle biotech and its plans for cutting-edge CAR-T cancer therapy. A year later, Juno is finally explaining what happened and promising to return with a new product.
The report, presented today at the Society for Immunotherapy of Cancer conference near Washington, DC, explains why the study participants, all with advanced acute lymphoblastic leukemia (ALL) that had resisted other treatments, succumbed unexpectedly to severe brain swelling, known as cerebral edema.
The report also sheds some light on Juno’s (NASDAQ: JUNO) decision last July to restart the study, dubbed ROCKET, after the first three deaths. A few months later, two more participants died and the study was shut down for good. The CAR-T product involved was shelved earlier this year.
CAR-T therapy has shown great promise, knocking out otherwise untreatable cancer, at least temporarily, in dozens of adults and kids with certain types of leukemia and lymphoma. The most promising form of CAR-T uses a patient’s own T cells, drawn from a blood sample, genetically modified outside the body, then re-introduced to the patient.
But it also has severe side effects. In some cases, like an overheated immune reaction called cytokine release syndrome (CRS), doctors are learning how to anticipate and control the worst symptoms.
CAR-T related cerebral edema, however, is more perplexing. There have been cases in other CAR-T studies, but nothing like the deadly cluster in Juno’s trial.
“It’s great they’re getting out in public with this,” says Stefan Grupp, head of cancer immunotherapy at Children’s Hospital of Philadelphia who treated the first pediatric patient ever with CAR-T.
With the deaths in the ROCKET study, Grupp says, the specter of CAR-T-related cerebral edema “went suddenly from ‘on no one’s radar screen’ to ‘major problem.’ It was really unexpected.”
Grupp is unaffiliated with Juno. His pioneering CAR-T work has been with CTL-019, a therapy that began at the University of Pennsylvania’s medical school, where he is a professor. In the hands of Novartis, it became the first CAR-T ever approved under the name tisagenlecleucel (Kymriah).
Juno officials have speculated for months that several factors contributed to the ROCKET deaths. Grupp says the study adds solid data to the theories, and one in particular: that the genetically modified cells, once pumped back into the patient, start to multiply too fast. “Everyone had this guess without data to support it that the speed of cell proliferation was an important factor,” Grupp says. “Their analysis shows that’s probably correct.”
The problem of cell proliferation is also present with CRS. When a patient has a lot of cancerous cells—a high tumor burden—the modified CAR-T cells rapidly multiply and attack, like sharks in a feeding frenzy, but also release a flood of damaging chemicals. In CRS, much of the damaging inflammation comes from the cytokine IL-6.
With cerebral edema, Juno officials say a key chemical to measure in CAR-T patients receiving CAR-T treatment is another inflammatory cytokine called IL-15. In the patients who died, they saw higher than typical IL-15 levels before infusion. Those IL-15 numbers spiked even higher after treatment.
Monitoring IL-15 levels could help identify new patients at higher risk before they receive their dose of CAR-T cells, says Juno chief medical officer Mark Gilbert. However, Juno will not exclude people from joining CAR-T trials simply on that basis. “It doesn’t change how we recruit and enroll,” Gilbert says, but it could help the company alert doctors about a patient’s status, or to choose whom to treat in a hospital setting instead of an outpatient clinic.
Beyond IL-15 levels, Juno identified other factors in the affected patients, such as low levels of blood platelets, which help with coagulation, and a particular genetic mutation.
The blood platelet information could be useful in clinical practice, says Richard Maziarz, a blood cancer specialist at Oregon Health & Science University in Portland. With many patients, doctors are willing to allow low counts, says Maziarz. “But maybe you need a higher platelet count for sicker patients with more disease burden.”
Juno Moves On
That said, JCAR015, the CAR-T product studied in ROCKET, has been retired. Juno’s new main CAR-T, called JCAR017, is very different. For example, JCAR015 used T cells straight from a patient’s own blood donation. JCAR017 is produced with a one-to-one ratio of two kinds of T cells: those that display a protein called CD8, which are particularly effective killers of viruses, bacteria, and cancer cells, and those that display the CD4 protein, which play more of a “helper” role.
Gilbert says that mix is critical. An uncontrolled amount of CD8-positive cells seem correlated with severe reactions. Those who died in ROCKET were all young—one was 40 years old, the rest were under 25—and not exposed to as many previous cancer treatments. With healthier immune systems, their own T cells, already modified to be cancer killers, in essence became too powerful. Pumped back into their bodies, the cells multiplied too fast. Perhaps because of their inflammatory secretions, the barrier that keeps fluid from leaking into brain tissue broke down.
With the T cell mix under more control with JCAR017, Juno thinks it has a better solution not just to the cerebral edema problem, but to other side effects as well. In patients with a type of non-Hodgkin Lymphoma, JCAR017 has produced relatively few safety problems: 44 of 69 in an ongoing study have had no CRS, cerebral edema, or other neurological side effects.
Gilbert says Juno will make its way back to adult ALL with JCAR017, but he declined to discuss when new clinical trials might begin. Maziarz, who treated NHL patients with Novartis’s CAR-T as part of a key study, is not affiliated with Juno. He is non-committal about Juno’s approach: “They’re controlling the variable with a formula that’s reproducible. Is it going to make a difference? That’s what they’re selling. I don’t know if it’s going to work.”
Last year, JCAR015 was one of three CAR-T programs racing to become the first-ever approved. The other two, from Kite Pharma and Novartis, have since received FDA approval. Kite’s treatment, axicabtagene ciloleucel (Yescarta), is now owned by Gilead Sciences, which acquired Kite for $11.9 billion in August.
When Juno announced the first three edema deaths in the ROCKET trial last July, officials pointed to one of two chemotherapies meant to prepare patients for the T cell treatment. The FDA allowed ROCKET to take off again without the chemo drug, fludarabine. It didn’t help. Two more patients died from edema without fludarabine.
Asked how the findings in today’s presentation shed light on Juno’s blame of fludarabine last year, Gilbert said: “We have always believed that multiple variables were at play.” But he notes that fludarabine seemed the most likely contributor because of the timing.
Gilbert says it’s easy in hindsight to say the high IL-15 levels could have been a warning, but in the earlier days of ROCKET, the pool of patients wasn’t large enough. “We went on to treat more patients, and the association is much stronger in the data set today,” he says. What’s more, Gilbert says, there’s still no real-time test for such underlying markers. “It takes time to do the analysis” from each patient. “While monitoring them, the fact is we didn’t have those data in hand.”
For Grupp, who treats children with ALL, there isn’t much in the Juno data to make him consider a change in clinical practice. He and his peers already know that patients with higher amounts of tumor cells are at higher risk of CAR-T side effects. The push, he says, is to get the treatment to the patients earlier in the course of disease.
T cell courtesy of NIAID via a Creative commons license.