CAR-T · Safety & follow-up
Acute toxicity: distinguish the syndromes
Fever, organ deterioration and neurologic symptoms need assessment—not an assumption that treatment is “working.”
In this chapter
- A safety map, not a bedside order set
- Safety disclosures through October 2026: serious signals, distinct programs
- What the disclosures establish—and what they do not
- Do not confuse LICATS with reassurance
- A clinician’s first distinction: treat the unstable patient, not the label
- Why a rare serious event can change a development program
A safety map, not a bedside order set
| Problem | Typical clinical concern | Interpretive priority |
|---|---|---|
| CRS | Fever with potentially escalating hypotension and/or hypoxia. | Grade and manage through a cellular-therapy protocol; infection may coexist. |
| ICANS | New confusion, language or handwriting change, altered consciousness or seizures. | Urgent neurologic and cellular-therapy evaluation; not every symptom is anxiety or fatigue. |
| IEC-HS | Hyperinflammation with evolving cytopenias, liver/coagulation abnormalities and marked ferritin changes. | Can arise as CRS resolves; requires syndrome-specific assessment, not a relabeling of every severe CRS. |
| ICAHT / cytopenias | Early, prolonged or recurrent reductions in blood-cell counts. | Consider conditioning, inflammation, infection and other causes; assess consequences, not only the count. |
| Infection / sepsis | Fever, instability or organ dysfunction at any relevant stage. | Do not delay infectious evaluation because CRS is plausible. |
| LICATS | Organ-local inflammatory changes in previously affected tissues after treatment. | Distinguish from active disease, infection, systemic toxicity and other organ injury. |
Consensus frameworks address systemic inflammatory and neurologic toxicity; the autoimmune literature has additionally described LICATS. These are distinct labels with overlapping possible manifestations, not interchangeable severity grades. [20][21][24]
Safety disclosures through October 2026: serious signals, distinct programs
From 24 August 2026, Novartis suspended screening, randomization and dosing in eight immunology and neuroscience studies of rap-cel following three reported fatal cases of immune effector cell-associated hemophagocytic lymphohistiocytosis-like syndrome (IEC-HS). The disclosed interruption did not include oncology studies. These are serious program-specific events; a pooled risk estimate across CAR-T products cannot be inferred from them. [29]
Bristol Myers Squibb (BMS) separately paused enrollment in autoimmune zola-cel studies after a concerning ICANS signal, which the company linked to a change in a manufacturing reagent. On 23 October 2026, BMS reported a death in the phase 1 Breakfree-2 multiple-sclerosis study: following late-onset motor symptoms not classified by the sponsor as ICANS, a participant received high-dose cyclophosphamide, developed septic pneumonia and died 44 days after CAR-T infusion. The causal contribution of zola-cel remains uncertain. BMS reported no other grade 5 events among 190 phase 1 autoimmune participants across its programs and was working with FDA toward a possible study restart in early 2027. BMS safety disclosure, 23 October 2026. [30][44]
What these events establish—and what remains uncertain
These sponsor disclosures are not independently adjudicated cross-product risk estimates. The Novartis IEC-HS fatalities and the BMS death following septic pneumonia arose in different studies and should not be combined into a single mortality rate. The BMS manufacturing-reagent association is the company’s reported explanation for an increased ICANS signal, not proof that a specific production method explains toxicity across platforms. Severe events warrant product-specific vigilance, updated consent and independent follow-up; they do not erase early evidence of durable benefit in carefully selected patients. Trial enrollment or restart must be verified directly with the sponsor and study site.
Do not confuse LICATS with reassurance
An organ-local post-treatment reaction is a diagnosis requiring clinical assessment, not a reason to observe worsening kidney, lung, cardiac or neurologic function without investigation. Apparent disease activity after treatment needs a differential diagnosis. Likewise, an inflammatory reaction is not a validated marker of therapeutic success. [20]
A clinician’s first distinction: treat the unstable patient, not the label
Consider fever and hypotension after infusion. A plausible inflammatory mechanism does not remove the possibility of infection. Consider recurrent fever with worsening blood counts and liver abnormalities after apparent CRS improvement. A different inflammatory syndrome may need consideration, rather than automatic reuse of the first diagnosis. Consider new focal weakness or altered consciousness. Neurologic toxicity, infection, vascular disease and other causes remain clinical questions. These examples are differential-diagnosis prompts, not diagnostic criteria.
The safe system is designed before the infusion: rapid access to the cellular-therapy team, an escalation plan, organ-specific expertise, and a clear handoff to local clinicians. These complications require urgent specialist assessment; patients should not try to manage possible CAR-T toxicities at home using a simplified self-treatment algorithm. [21][23][24]
Why a rare serious event can change a development program
A small study with no severe event can still be compatible with a clinically important uncommon risk. Conversely, a safety signal in one program does not mathematically establish the same incidence in all products. The useful next questions are the denominator, exposure, timing, affected phenotype, adjudication, concomitant treatment and any revised risk-mitigation plan. Until those details are available, numerical reassurance and numerical alarm can both be misleading.
Sources · 7
Source titles are shown in their original language.
[20] Primary research · 2025-04-30
Hagen et al. Local immune effector cell-associated toxicity syndrome in CAR T-cell treated patients with autoimmune disease: an observational study. Lancet Rheumatology.
Open original source ↗[21] Expert consensus · 2023-03-09
Hines et al. Immune Effector Cell-Associated Hemophagocytic Lymphohistiocytosis-Like Syndrome. Transplant Cell Ther.
Read with the published erratum; clinical algorithms must use corrected units.
Open original source ↗[23] Expert consensus · 2024-07-30
Shahid et al. ASTCT best practice: infection prevention and management after CAR T-cell therapy for hematological malignancies.
Oncology-based recommendations; adapt within autoimmune trial/center protocols, not automatic identical application.
Open original source ↗[24] Expert clinical framework · 2025-11-28
Caricchio et al. A Guide for Initiating and Managing CAR T Cell Therapy Clinical Trials in Autoimmune Rheumatic Diseases. ACR Open Rheumatology.
Open original source ↗[29] Original company-spokesperson reporting · 2026-09-01
FirstWord Pharma. Novartis pauses immunology/neuro CAR-T studies after three patient deaths. Direct company-spokesperson reporting.
Public safety disclosure, not a peer-reviewed denominator-based safety analysis; no verified restart identified in this review.
Open original source ↗[30] Original news reporting · 2026-09-01
Reuters. Novartis pauses autoimmune cell-therapy studies; separate BMS zola-cel pause reported.
Used to distinguish BMS reversible inflammatory events from the Novartis fatalities. Do not combine the two safety records.
Open original source ↗[44] Original reporting with direct company-spokesperson statements · 2026-09-01
BioSpace. Novartis pauses CAR T studies due to 3 deaths; BMS follows suit after safety events.
A safety disclosure, not an adjudicated trial analysis or a denominator-based estimate.
Open original source ↗