Interpreting CAR-T safety: lessons from zola-cel so far Save

While there is plenty to watch this ACR in the maturing CAR-T space, one of the big global stories in the second half of 2026 has been about zola-cel and the pause in the Breakfree program. 

The latest SLE data from the full phase 1 cohort (which also included scleroderma and myositis patients) will be presented as a late-breaking poster on Tuesday (LB04) and it will report excellent efficacy, with 20 of 22 previously treatment-refractory SLE patients were controlled without going back on their conventional lupus therapies, and phase 2 started in July 2025. Just like at the multiple sclerosis meeting a few weeks ago, there will be enormous scrutiny, however. The high-stakes label that often gets applied to CAR-T also applies to the lay media reporting, and it tells you a lot about the wild ride we’re going to have in interpreting safety in CAR-T.

And thinking about how to interpret safety is important - because perhaps the two biggest questions about CAR-T for the lay rheumatologist on the streets of Orlando are (a) is this too big a risk for our patients, with direct implications for patient selection (b) how hard will this be to deliver when manufacture is included, with direct implications for cost, and therefore patient selection.

Of course, the questions about efficacy (including depth, universality, and durability) still need answering, not all rheumatic diseases are due to an excess of dysfunctional B cells, and even for the diseases which are no doubt even a safe and cheap CAR-T is not going to be for everyone. Given that the upside hype doesn’t really need any assistance, the real questions, though, are about safety and logistics/cost.

If you’ve been following the CAR-T story so far, you’ll know how important rapid manufacture is, especially if you appreciate a bit of rheumatic disease therapy history - because once upon a time TNF inhibitors were beset by unwieldy manufacture and it was thought lengthy inpatient stays would be needed for administration. Conventional CAR-T has involved harvesting patient T cells, engineering them, expanding them extensively in a factory, and then infusion a large number of relatively differentiated CAR-T cells back into the patient. If you’ve wondered where the price tag comes from, it’s from this.

Zola-cel is one of a few products with a rapid manufacturer autologous approach. In short, the collection and transduction still occurs ex vivo, but the major expansion occurs inside the patient with relatively immature stem-like CAR-T. This means zola-cel (BMS) spends less than five days in manufacturing using its NEX-T platform, and the rival (and analogous but separate) rap-cel (Novartis) spends just two days using the T-Charge platform. Compared to allogeneic CAR-T and in vivo CAR, this approach is still quite conservative, but it’s a game changer next to conventional CAR-T, and also the most obvious next step to make CAR-T achievable on scale.

On August 31, Endpoint News broke a headline “Three deaths in CAR-T trials prompt halt by Novartis; Bristol Myers programs also paused”, followed by the Wall Street Journal, and then Reuters and many other outlets. There had been three episodes of fatal immune effector cell-associated haemophagocytic syndrome (IEC-HS) with rap-cel, despite initially reporting reassuring safety data. There is still much we don’t know about this, but the easy conclusion had been that the explosive in vivo expansion comes at an inflammatory toxicity risk. When almost at the same time it surfaced that the zola-cel ‘Breakfree’ program had already been paused for two months on safety grounds, it was easy to question whether this whole approach was coming to a halt.

The details matter, though, and have played out over the last two months. It was clarified the zola-cel voluntary pause was for ICANS (immune effector cell-associated neurotoxicity syndrome, a known complication of CAR-T) in the phase 2 Breakfree-SLE study, and that these were reversible without clinical sequelae. ICANS had also similarly occurred and been reversed in the phase 1 SLE study, as well as in the phase 1 MS study, including one life-threatening episode that resolved with anakinra and steroids. The ICANS in the phase 2 Breakfree-SLE study was more frequent, but still reversible, and subsequently the problem has been attributed by the manufacturer to a manufacturing change involving a reagent. While the details have not been made public, the manufacturer reports making adjustments and plans to restart.

There was a death, though - in the phase 1 MS trial, for a non-ICANS cause with delayed motor symptoms, and further immunosuppression was followed by fatal septic pneumonia, 44 days after infusion, occurring after the program was paused. We don’t know why this happened or whether it is MS specific, but it was definitely discussed at length at the MSToronto (joint North American-European) meeting two weeks ago.

It would seem, however, that the bigger safety risk here is arguably from lymphodepletion. This is slightly uncomfortable given that the biggest attraction of CAR-T in SLE is escaping chronic immunosuppression - we hope we can get around this, but currently this might be the Shawshank Redemption-style five hundred yards of crawling to freedom to truly break free.

So a few points to reflect on here. 

First, problems with one program do not necessarily doom other similar but separate programs, and we will need to consider each safety situation judiciously rather than CAR-T as a monolith. Secondly, the value in pursuing rapid manufacture is clearly alluring enough that we are not (and should not) be completely discouraged at the first sign of bad weather. Thirdly, as this gets tested quickly across indications, it’s unclear what is specific to the underlying disease versus what isn’t. And finally, given a lot of risk still sits with lymphodepletion, we will watch the immunosuppression-light in vivo approaches very closely, even though they have their own issues.

It’s a fascinating time for drug safety, especially as we watch all that happens with CAR-T at this meeting and beyond. Predicting how the CAR-T story plays out will rely on understanding the intricacies of safety.