Novartis Pauses CAR-T Autoimmune Trials After Three Patient Deaths; Bristol Myers Follows

Novartis (NYSE: NVS) temporarily halted eight clinical trials of rap-cel, its experimental CAR-T cell therapy for autoimmune and neurological disorders, after three patients died from a severe immune response, The Wall Street Journal reported on September 1. Bristol Myers Squibb (NYSE: BMY) paused its own CAR-T trials in autoimmune disease as a precaution, the same report said. NVS closed at $152.06 on August 31, down 1.1% from Friday’s $153.81; BMY closed at $66.81, up 0.3%. The news broke before Tuesday’s open, so the first full market reaction is still ahead.

rap-cel engineers a patient’s own T cells to attack a specific target, and Novartis is testing it in lupus, multiple sclerosis, and other autoimmune and neurological conditions. The three deaths stemmed from a severe immune reaction, a recognized CAR-T risk most often tied to cytokine release syndrome or immune effector cell-associated neurotoxicity. Novartis has not said when the patients died or specified the mechanism. The company told the Journal it is running “a comprehensive review of the observed safety events,” working with the independent safety boards that oversee each paused study and sharing what it knows with regulators.

Why It Matters

CAR-T’s move from oncology into autoimmune disease is one of biotech’s biggest bets. Novartis, Bristol Myers, Kyverna, Cabaletta, and others are racing to prove the technology can reset the immune system in lupus, MS, and other chronic conditions. Novartis itself launched the first approved CAR-T, Kymriah, back in 2017 for blood cancer. But autoimmune patients are generally healthier than the end-stage cancer patients CAR-T was originally built for, which raises the bar: three deaths from a severe immune reaction land differently when baseline mortality is lower. This is the most significant safety signal yet for CAR-T in autoimmune disease and could slow the entire field’s expansion into chronic indications. For how the platform works, see our CAR-T explainer.

What Happens Next

The read-through is sharpest for pure-play CAR-T autoimmune developers. Cabaletta (Nasdaq: CABA) closed at $3.11 on August 31; Kyverna and others run similar programs in lupus. Three things to watch: details on which trials and doses the deaths occurred in, whether the FDA imposes a broader hold beyond the company-initiated pause, and whether other sponsors follow Bristol Myers’ lead. Novartis’s review will set the tone for how the whole field proceeds. We covered the investor setup in our CAR-T stocks piece.

What It Means for the Trade

Position this event correctly: it is a class-level risk-off trigger for autoimmune CAR-T, and the damage is proportional to how much of a company’s value sits in that single thesis. Novartis and BMS are diversified mega-caps where autoimmune CAR-T is a pipeline line-item — the 1% moves on the news reflect that. Cabaletta at $3.11 and Kyverna are the leveraged expressions: their entire enterprise value rests on CD19-directed CAR-T succeeding in lupus, and three deaths on a competitor’s program raise fundamental questions about whether the risk-benefit math that works in refractory oncology survives translation into chronic autoimmune disease.

The specific question the deaths force: what went wrong mechanistically. “Severe immune reaction” consistent with CRS or ICANS is the standard CAR-T toxicity profile, but the oncology experience says those risks are largely manageable with tocilizumab, steroids, and experienced centers — mortality from CRS/ICANS in lymphoma patients runs well under 5% at seasoned sites. If three autoimmune patients died despite that playbook, the bear interpretation is that something about the autoimmune context (different lymphodepletion tolerance, healthier immune systems mounting stronger cytokine responses, or rap-cel’s specific construct) amplifies the risk. If instead the deaths trace to protocol deviations or patient selection, the class-wide damage is containable. The market cannot distinguish these scenarios yet — which is why every autoimmune CAR-T name trades with a discount until Novartis discloses more.

Also note what the event does to the competitive hierarchy. Deep-phenotype alternatives to full CD19 CAR-T — telitacicept-style B-cell modulation, BCMA CAR-T with different trafficking, and notably the bispecific antibody approach (incidentally the class Amgen’s Imdelltra validated in oncology, now being extended toward autoimmune) — just got relatively more attractive to investors who want autoimmune exposure without cell-therapy logistics risk. Money rotating out of CAR-T pure-plays doesn’t leave the theme; it moves one seat over.

Field Context: What Was Being Built

The stakes are worth spelling out because the bull case was genuinely large. The concept: autoimmune diseases like lupus are driven by a self-replenishing pool of autoreactive B cells and plasmablasts; wipe them out with a one-time CD19-directed CAR-T (the same Kymriah-style construct that transformed B-cell malignancies) and the immune system “resets” — early trials from German academic groups showed drug-free remission in refractory SLE patients lasting years after a single infusion. The commercial pitch was a potential functional cure for diseases currently managed with lifelong chronic immunosuppression — hence Novartis launching eight trials across lupus, MS, and other conditions, BMS pointing its Celgene-derived cell-therapy machine at the same space, and venture money flooding Kyverna, Cabaletta, and Cartography.

The unresolved tension was always safety economics. In lymphoma, a 3-5% serious-toxicity rate is acceptable against a lethal disease. In a 25-year-old lupus patient whose disease is controlled (imperfectly) by standard agents, the acceptable mortality threshold is closer to zero. That’s the real story of this pause — not that CAR-T stopped working, but that the field’s risk tolerance may be lower than its biology requires.

Risks and What Could Go Wrong

For the class: an FDA-imposed class-wide hold or new safety requirements (stricter site certification, mandatory ICANS monitoring protocols, longer inpatient stays) would raise development costs across every autoimmune CAR-T program and delay timelines by quarters — recoverable but painful, especially for cash-constrained small caps. For Novartis specifically: rap-cel was a pillar of its post-Entresto growth diversification, and a protracted review or program death joins a growing 2026 failure list (pelacarsen, del-desiran) that pressures the pipeline narrative. For the trade: the highest-beta expressions of an unfavorable resolution are the small caps, and the highest-beta expression of a favorable one (deaths explained, trials resume) is the same small caps — either way, position size for the binary, and don’t average into names whose value depends on a safety question nobody outside Novartis’s safety boards can currently answer.