Autoimmune CAR T cells move fast, but safety is now driving the pace
The early story of autoimmune CAR T therapy was built around dramatic rescue. A young woman with severe systemic lupus erythematosus entered medication-free remission after a single infusion of CD19-targeting CAR T cells, and that result helped propel a field that now includes more than 100 trials, mostly in phase 1 and 2. The latest clinical picture is more complicated. Regulators, trial teams, and investigators are now confronting the same therapy’s capacity to produce deep immune reset and serious toxicity in a population that is not facing immediate cancer death.
Two major pauses in late August made the trade-off hard to ignore. Novartis halted multiple autoimmune trials of rapcabtagene autoleucel after deaths associated with immune effector cell-associated hemophagocytic syndrome were reported, while Bristol Myers Squibb paused enrollment and treatment in studies of zolacabtagene autoleucel after seeing a higher frequency of ICANS, or immune effector cell-associated neurotoxicity syndrome, in a phase 2 study. In oncology, those events would already be serious. In chronic autoimmune disease, where patients may live for years on established therapies, they raise a different threshold question: how much risk is acceptable for a treatment that aims to create long-term remission rather than short-term stabilization?
Remission signals remain strong in severe lupus and related disorders
The strongest clinical signal still comes from B-cell–directed therapy. In the Erlangen phase 1/2 basket trial, 24 adults with severe autoimmune disease, including lupus and systemic sclerosis, received a single infusion of Zorpo-cel, a CD19-targeting CAR T cell therapy. Twenty-two participants achieved drug-free disease control for at least 24 weeks. That kind of response is biologically plausible because CD19-directed CAR T cells can deplete pathogenic B cells more completely than conventional immunosuppressive regimens, potentially interrupting autoantibody production, antigen presentation, and downstream inflammatory amplification at the same time.
The absence of the classic CAR T toxicities was striking. No cases of severe cytokine-release syndrome and no ICANS appeared in the Erlangen trial. Carl June called that another surprise, and the reaction is understandable: the field still carries the memory of early oncology experience, when systemic inflammation and neurotoxicity set the terms of the safety debate. Here, the expected toxicity pattern did not materialize, at least in this small study. A single infusion produced durable disease control without the inflammatory collapse that many clinicians had braced for.
A broader evidence base is starting to take shape
The Erlangen experience no longer stands alone. A 2025 pooled analysis of 145 participants across 16 studies found that 70% of patients with lupus reached remission. Serious adverse events were uncommon but not absent: one participant developed cytokine-release syndrome, two developed ICANS, and 11 patients, or 7.6%, had serious infections, including one fatal pneumococcal meningitis. Those numbers matter because they frame the therapy’s promise and its price. Even when the acute CAR T toxicities are infrequent, profound immune suppression can create a downstream infection burden that may be less visible than neurotoxicity but no less consequential at the bedside.
A European case series in 2026 extended the signal into pediatric disease. Eight participants, four with childhood-onset lupus, received CD19-targeting CAR T cells and showed clinically substantial improvement or disease resolution for a median of 16.5 months. No severe adverse events were reported in that series. Jackson, who is running the REACT-01 phase 1 trial in children with lupus, argues that earlier treatment may allow clinicians to intervene before organ damage becomes fixed. For children with severe disease, the logic is hard to dismiss: if B-cell reprogramming can arrest inflammation before cumulative injury accumulates, the clinical benefit could be larger than in adults with long-standing damage.
The risk calculus changes outside refractory disease
The current debate is not centered on whether CAR T can work. It is centered on when the risk is justified. For patients with no other viable options, most experts appear willing to accept the burdens of manufacturing, lymphodepleting chemotherapy, and prolonged follow-up. For patients whose lupus or other autoimmune disease remains controllable with standard drugs, the calculation becomes far less forgiving.
Coziana Ciurtin has pushed that point directly. She noted that many patients with lupus can manage symptoms on existing medication, and that remission lasting months or even a few years is not unheard of in routine lupus care. Her warning is practical: clinicians have to be transparent about the experimental nature of CAR T, the known and uncertain risks, and the established alternatives already available. A single-cell infusion may be easier to explain in a rescue setting than in a patient who is functioning reasonably well on conventional therapy but wants a shot at prolonged remission.
The treatment burden itself is substantial. Standard CAR T delivery requires cell collection, individualized manufacturing, a short course of lymphodepleting chemotherapy, and then infusion. Some patients may require 15 years of safety monitoring. Those logistics are manageable when a disease is life-threatening and refractory. They are harder to justify when the same disease can be controlled with safer treatments, even if those treatments are not curative.
Trial pauses sharpen the field’s next safety questions
The recent suspensions by Novartis and Bristol Myers Squibb have forced a more sober view of the platform. Nirali Shah, who helped define immune effector cell-associated hemophagocytic syndrome, said the events should not dissuade the field from learning from them and moving forward. Saurabh Mehandru framed the problem in even starker terms: the field now has to understand toxicity mechanisms and determine how to treat those toxicities when they occur.
That is the next clinical frontier. CAR T is not a single intervention but a collection of variable constructs, targets, lymphodepletion regimens, manufacturing methods, and follow-up protocols. Some details remain commercially confidential, which limits direct comparisons across studies and weakens any attempt to generalize from a small trial to routine care. In autoimmune disease, where the patient population is broader and the acceptable risk threshold is lower than in oncology, those details are not academic. They define who should be treated, how aggressively, and under what monitoring safeguards.
The field now has enough signal to know that CD19-targeting CAR T can induce deep remissions in severe autoimmune disease. It also has enough warning signs to know that the same platform can produce fatal inflammatory complications and neurotoxicity in other programs. The unresolved issue is not efficacy in the abstract. It is whether a therapy designed to erase pathologic immunity can be made safe enough for diseases that are serious, chronic, and often already manageable with less hazardous treatment.