Juno analysis of shuttered study offers clues for CAR-T
- Juno Therapeutics Inc. on Friday offered up new details about what factors might have contributed to the deaths of five patients last year in a clinical study of its now-discontinued CAR-T candidate JCAR015, identifying several product and patient characteristics connected with higher risk of deadly neurotoxicity.
- Early and rapid expansion of CAR T-cells in the body after infusion seem to correlate with severe brain toxicity in patients treated with the therapy. That amplification of modified T-cells, in turn, appeared partially linked to JCAR015's variable cell composition, along with certain patient attributes.
- CAR-T remains a new technology, despite the field's rapid advance. Currently marketed therapies from Novartis AG and Gilead Sciences Inc. are only approved for a small number of heavily pre-treated patients. Moving CAR-T treatment earlier will require an improved understanding of the potential for severe side effects.
Since the patient deaths derailed Juno's ROCKET study, the Seattle, Washington-based biotech has taken a back seat to rival CAR-T developers. Figuring out what went wrong with JCAR015 has been an important step in Juno's efforts to recover and move forward.
When Juno reported the first three patient deaths in its ROCKET study last summer, CAR-T therapy's risk/benefit balance was less certain. The cerebral edema, or severe brain swelling, which led to the deaths sparked safety concerns for the entire field — particularly as it wasn't clear what exactly had caused the fatal reactions.
Juno's first attempt to address the issue only intensified those worries. After the Food and Drug Administration put the ROCKET study on hold, Juno proposed removing a chemotherapy agent used to precondition patients before they received an infusion of CAR-T cells.
The company's early hypothesis linked the combination of this drug — called fludarabine — with the modified immune cells as a possible cause. But after the FDA OK'd a trial restart, two more patients died from cerebral edema. Across all 38 trial participants evaluable for safety from ROCKET, Juno reported severe neurotoxicity in 52% of patients.
Both neurotoxicity and an immune reaction known as cytokine release syndrome are common side effects to CAR-T treatment. But the rate of fatal cerebral edema in ROCKET was much higher than other studies.
Juno eventually shut down the ROCKET study entirely and discontinued JCAR015, turning instead to a differently constructed candidate known as JCAR017.
Hunting for answers
Since last year, Juno has been analyzing data from ROCKET to track down root causes for the deaths.
What has become clear from Juno's analysis is that the answer is less likely a single trigger, but rather a web of interconnected variables. While that makes solving the questions posed by ROCKET more difficult, it underscores the complexity of CAR-T.
Juno linked rapid expansion of CAR-T cells following infusion with a higher risk of neurotoxicity, although the process by which this early proliferation leads to more severe reactions still is not fully understood.
"The exact mechanism is harder to come by," explained Mark Gilbert, Juno's chief medical officer, in an interview. "It will be important to understand better, because by understanding it you have the ability to use either preventative therapies or treatments against it."
"But we are still a little short of that," he cautioned.
Higher risk was also associated with increased expression of an inflammatory biomarker known as IL-15, which Juno identified as a CAR-T growth factor.
Perhaps more intriguingly, patients whose tumors expressed a non-Philadelphia type gene expression signature experienced greater toxicity. Among 15 patients with this type of genetic signature, 12 had Grade 3 or higher neurotoxicity, including all five patients who died.
This result is surprising, says Gilbert, and suggests the relationship between CAR-T cells and tumor cells expressing the relevant target may be more complex than currently thought.
"You have your CAR-T cell. It recognizes CD19. If you have a cell that has CD19 on the surface, it kills it. End of story. That was the paradigm under which we operated," Gilbert explained.
"That may still be true, but there may be other nuances that are coming in where certain types of cells actually modify the activity of the T-cell or they may provide benefits to other parts of the body."
These correlations, while exploratory and not backed by statistical powering, point to possible areas drugmakers and researchers can explore to better understand which patients are at higher risk from CAR-T therapy. Better patient stratification could be particularly important as the field pushes into earlier lines of treatment — where the tradeoffs on safety are much different.
Indeed, patients with two or fewer prior treatment regimens were at higher risk of neurotoxicity, according to Juno's analysis.
Taken as a whole, Juno's analysis underscores how the CAR-T field is still developing. Both Novartis and Gilead have notched major successes in securing U.S. approvals for their respective therapies, each of which have delivered remarkable results in some patients. Promising results from bluebird bio, Nanjing Legend Biotech and others are proof of broader utility. Like immuno-oncology at large, though, CAR-T drugmakers are still learning what works.
Follow Ned Pagliarulo on Twitter