There is a particular kind of phone call an oncology ward learns to dread. The patient was steady at breakfast, and by afternoon cannot hold a sentence together or track the date, and within hours a few of them are seizing. This is ICANS, immune effector cell-associated neurotoxicity syndrome, and it is the price a subset of cancer patients pay for the engineered T-cells that were supposed to save them. Nobody at the bedside narrates it as a tradeoff. They page neurology and reach for the dexamethasone.
So when UC Irvine announced on August 18 that its researchers had “developed a framework to make CAR T-cell therapy safer,” it landed on a real and frightening problem. The framing invites you to picture a lab, a result, a thing that now exists that did not before. Read the actual paper in Frontiers in Pharmacology and a more modest object comes into focus. It is a Perspective article. It contains no new experimental data. The work proposes. It does not test.
It also proposes something with a particular commercial shape, which the university’s summary left out. The paper’s answer to CAR-T neurotoxicity is not to swap out the toxic therapy but to keep patients on it and layer a second drug on top to manage the damage. And its corresponding author, medical oncologist Erin A. Dean, consults for Kite Pharma, the company that makes the cell therapy with the worst neurotoxicity record on the market. Hold that thought; the record is coming.
The argument itself is a good one, which is worth saying before taking it apart. Lead author Atena Zahedi and her colleagues propose that the brain injury of CAR-T is not simply collateral damage from cytokines sloshing across the blood-brain barrier, but something that takes on a life of its own inside the neuron’s mitochondria. Cytokine stress cripples the electron transport chain, reactive oxygen species pile up, the NLRP3 inflammasome and the cGAS-STING pathway switch on, and the whole thing becomes what the authors call a feedforward loop that sustains neuroinflammation even after the systemic cytokine surge has cooled. If that model holds, steroids are treating the smoke and missing the fire.
Their proposed extinguisher has a satisfying shape. For a toxicity thrown off by cell therapies that carry six-figure price tags, the fix on offer might be an old pill that costs a few dollars: leflunomide and its active metabolite teriflunomide, borrowed from multiple sclerosis, and metformin, the diabetes generic. The pitch is repurposing, which skips years of development and, for the approved generics at least, rests on a long safety record. The list runs further, into the mitochondrial antioxidant elamipretide and experimental NLRP3 inhibitors, which are neither cheap nor established, but the core of it is a shelf of drugs the pharmacy already stocks.
A satisfying inversion is not a finding, though, and here the distance between the press release and the paper matters. The announcement says the researchers “developed a framework.” The paper describes a staged pathway that still has to travel, in the authors’ own words, from preclinical validation to a randomized clinical trial before anyone gives it to a patient. The authors are honest about the hardest part: inhibiting DHODH to calm the brain could also blunt the CAR T-cells’ expansion and therefore their ability to kill the cancer, and they float a “pharmacological rescue” to patch that hole. The mechanism you would suppress to protect the brain is one the therapy may need to work. That tension does not appear anywhere in the word “framework.”
The scale of the problem is not in dispute, and it is where the incentive comes into focus. A 2024 meta-analysis in Frontiers in Neurology pooled 75 trials and 3,184 patients and put all-grade ICANS at 26.9 percent and severe cases at 10.5 percent, with the CD19-targeted products higher at 34.8 percent all-grade and 14.7 percent severe. Axicabtagene ciloleucel, the CD28-based product sold as Yescarta, ran higher than either, at 54.0 percent all-grade. The same review counted 6 deaths from ICANS, grade 5, in real-world use, and every one was on Yescarta.
That is the therapy Dean’s consulting client makes. None of it is buried; the disclosure sits in the competing-interests statement where it belongs. It simply reframes the story. An academic who consults for the maker of the cell therapy with the worst neurotoxicity record is co-authoring a paper on how to keep patients on cell therapy by adding a drug to manage that neurotoxicity. That is not a scandal, it is an incentive, and it is the kind of thing a reader deserves to see named rather than airbrushed out of a university communications office’s summary.
To their credit, the authors took no money for the work; the paper states plainly that no financial support was received. The science may well prove out. Mitochondrial dysfunction is a legitimate target, and a cheap repurposed drug that spared some fraction of patients a week of confusion and seizures would be a gift worth having.
But that is a trial yet to be run, not a framework already built. Zahedi and Dean wrote a hypothesis and labeled it one. The promotion to breakthrough happened a floor or two up, in the office that turned an essay into a framework and a framework into a headline.
Sources
- Frontiers in Pharmacology – Zahedi et al., Perspective on mitochondrial-targeting drugs for ICANS (2026; DOI 10.3389/fphar.2026.1896516), including funding and conflict-of-interest disclosures
- Frontiers in Neurology – meta-analysis of ICANS incidence across 75 trials / 3,184 patients (2024)
- News-Medical – “UC Irvine researchers develop framework to make CAR T-cell therapy safer” (Aug 18, 2026)