Roughly 5 to 10 percent of patients who receive CAR T-cell therapy die from causes unrelated to their cancer relapsing, a figure that varies depending on the specific disease being treated and the product used. A large meta-analysis covering more than 7,600 patients found treatment-related death rates ranging from about 6 percent for large B-cell lymphoma to nearly 11 percent for mantle-cell lymphoma. Those numbers make CAR-T one of the more aggressive treatments in oncology, but they exist in a context where the patients receiving it have already exhausted other options and face poor odds without it.
How the Numbers Break Down by Disease
The risk of dying from the treatment itself, rather than from cancer progression, is what researchers call nonrelapse mortality, or NRM. A 2024 systematic review and meta-analysis pooling data from 18 clinical trials and 28 real-world studies found that NRM point estimates differed substantially across cancers. Mantle-cell lymphoma carried the highest risk at about 10.6 percent, followed by multiple myeloma at 8.0 percent, large B-cell lymphoma at 6.1 percent, and indolent lymphoma at 5.7 percent.1Nature Medicine. A systematic review and meta-analysis of nonrelapse mortality after CAR T cell therapy Those percentages represent patients who survived long enough to receive their infusion but then died from a treatment complication, not from their underlying cancer.
Inpatient mortality, which captures deaths during the hospitalization for CAR-T infusion, has held relatively steady at around 3 percent in the United States even as the number of procedures climbed from 70 in 2017 to roughly 4,700 in 2022.2Transplantation and Cellular Therapy. Trends and Disparities in Chimeric Antigen Receptor T-Cell Therapy Utilization, Inpatient Mortality, Length of Stay, and Costs in the United States (2017 to 2022) The gap between that 3 percent inpatient figure and the higher NRM numbers reflects the fact that a significant share of treatment-related deaths happen after patients leave the hospital, sometimes months later.
What Actually Kills Patients After CAR-T
Infections are by far the leading cause of treatment-related death. Data from a large French registry of patients with large B-cell lymphoma found that among 48 patients who died from nonrelapse causes, infections accounted for 56 percent of those deaths, split roughly evenly between COVID-19 and non-COVID infections. Cytokine release syndrome caused about 10 percent, strokes and brain hemorrhages together accounted for another 12 percent, and second cancers made up 6 percent. Neurotoxicity and miscellaneous causes rounded out the remainder.3PubMed Central. Nonrelapse mortality after CAR T-cell therapy for large B-cell lymphoma: a LYSA study from the DESCAR-T registry
That breakdown is important because it shows that the dramatic, acute reactions people associate with CAR-T, like cytokine storms, are actually a minority of the lethal events. The quieter, slower threat of infection is what claims the most lives. This pattern holds across multiple datasets and disease categories.
Why Infections Are So Dangerous After CAR-T
Before CAR T cells are infused, patients receive a round of chemotherapy called lymphodepletion. This is designed to clear out the patient’s existing immune cells so the engineered T cells have room to expand. The result, by design, is a profoundly weakened immune system. Patients develop low white blood cell counts, depleted antibody levels, and a loss of B cells that can persist for months or longer.4PubMed Central. Infectious complications of chimeric antigen receptor (CAR) T-cell therapies On top of that, if the patient develops cytokine release syndrome or neurotoxicity, the drugs used to treat those complications, such as steroids and certain immune-blocking agents, deepen the immunosuppression further.
The types of infections shift over time. Bacterial infections tend to dominate in the early weeks after infusion, during the period when white blood cell counts are at their lowest. As weeks turn to months, viral reactivations and fungal infections become more prominent, especially in patients whose immune recovery is slow.5PubMed Central. Infectious complications after CAR T-cell therapy: mechanisms, risk stratification, and prevention Late infections are particularly concerning because they can strike patients who feel well and may have returned to normal life, catching both patients and care teams off guard.6PubMed Central. Managing Infection Complications in the Setting of Chimeric Antigen Receptor T cell (CAR-T) Therapy
Cytokine Release Syndrome and Brain Toxicity
Cytokine release syndrome, commonly called CRS, is the side effect most closely associated with CAR-T in the public imagination. It occurs when the engineered T cells activate en masse and trigger a flood of inflammatory signaling molecules. This can cause fevers, plummeting blood pressure, difficulty breathing, and organ damage.7PubMed Central. Mechanisms of cytokine release syndrome and neurotoxicity of CAR T-cell therapy and associated prevention and management strategies Most cases of CRS are now manageable with drugs like tocilizumab, which blocks one of the key inflammatory signals, and steroids. Fatal CRS still occurs but is relatively rare compared to the early days of the therapy.
Neurotoxicity is a related but distinct complication. Immune effector cell-associated neurotoxicity syndrome, or ICANS, can range from confusion and difficulty finding words to seizures and, in extreme cases, fatal brain swelling. The mechanism appears to involve inflammation that damages the blood-brain barrier, allowing immune signals to penetrate the brain and trigger swelling in brain cells. In the most severe scenario, this cycle feeds on itself, with worsening barrier damage leading to more swelling, culminating in fulminant cerebral edema.8PubMed Central. A Case of Fulminant Cerebral Edema Leading to Death After Chimeric Antigen Receptor T-Cell Therapy Fatal cerebral edema is rare, but case reports highlight how rapidly it can progress.
Beyond CRS and ICANS, other recognized toxicities include a condition resembling hemophagocytic lymphohistiocytosis, where the immune system goes into overdrive and begins destroying the patient’s own blood cells, and prolonged low blood counts that leave patients vulnerable for extended periods.9PubMed Central. Mechanisms and management of CAR T toxicity
Heart Complications
Cardiovascular toxicity has received increasing attention as more patients receive CAR-T. A review of the literature found that the most common cardiac events include dangerous drops in blood pressure requiring medications to support circulation, abnormal heart rhythms like atrial fibrillation, heart failure, new cardiomyopathy, myocarditis, and cardiac arrest.10PubMed Central. Cardiotoxic Effects Following CAR-T Cell Therapy: A Literature Review Many of these events are linked to the inflammatory cascade of CRS rather than being a direct effect of the CAR T cells on heart tissue. Patients with pre-existing heart conditions face a higher baseline risk, which is part of why careful cardiac screening before infusion matters.
When Deaths Occur After Infusion
Understanding the timing of treatment-related deaths helps both patients and clinicians plan for risk. An Italian prospective study tracking nearly a thousand patients found that about 40 percent of nonrelapse deaths occurred within the first 28 days after infusion, roughly 23 percent between days 29 and 90, and about 36 percent after the 90-day mark.11Transplantation and Cellular Therapy. Systemic Inflammation and CAR-T Specific Toxicities as Major Drivers of Nonrelapse Mortality: Analysis from the Italian Prospective Observational CART-SIE Study The French registry data tell a similar story: only about 19 percent of nonrelapse deaths happened in the first four weeks, with the vast majority, 81 percent, occurring later.3PubMed Central. Nonrelapse mortality after CAR T-cell therapy for large B-cell lymphoma: a LYSA study from the DESCAR-T registry
The fact that more than half of treatment-related deaths happen after the acute post-infusion window challenges the common assumption that the biggest danger is the initial cytokine storm. It underscores how much of the real risk is tied to the prolonged immune vulnerability that follows treatment.
Predicting Who Is at Highest Risk
Researchers have been working to identify patients most likely to develop dangerous complications before infusion even begins. Simple blood tests drawn before lymphodepletion show promise. A study found that combining a measure of endothelial stress with ferritin and C-reactive protein levels could sort patients into three risk groups. The highest-risk group had about a 74 percent chance of developing moderate-to-severe CRS and a 74 percent chance of significant neurotoxicity, while the lowest-risk group’s rates were roughly 23 percent and 29 percent, respectively.12Blood Advances. CRP and ferritin in addition to the EASIX score predict CAR-T–related toxicity
In multiple myeloma specifically, high pre-infusion ferritin levels have been linked to substantially worse survival. A meta-analysis found that patients with elevated ferritin had roughly triple the risk of dying compared to those with lower levels.13PubMed Central. Serum ferritin as a prognostic biomarker in CAR-T therapy for multiple myeloma: A meta-analysis Elevated ferritin generally signals higher underlying inflammation and greater disease burden, both of which make patients more fragile heading into an already demanding treatment.
These kinds of risk-stratification tools are not yet universally standardized, but they are already influencing clinical decisions: whether to preemptively start anti-inflammatory drugs, how long to keep a patient hospitalized, and whether outpatient monitoring is safe for a given individual.
Does Treating Side Effects Undermine the Therapy?
One persistent worry among patients and clinicians alike is that the drugs used to manage side effects, especially steroids and tocilizumab, might dampen the CAR T cells and reduce their ability to fight the cancer. This concern has contributed to a culture of sometimes delaying intervention for side effects, hoping to preserve the therapy’s potency. However, evidence from patients with multiple myeloma suggests that timely use of corticosteroids and tocilizumab to control immune-mediated toxicities does not appear to undermine the long-term antitumor effect of CAR-T.14Blood Cancer Journal. Prognostic impact of corticosteroid and tocilizumab use following chimeric antigen receptor T-cell therapy for multiple myeloma That finding has shifted practice toward earlier and more aggressive toxicity management, which itself may be contributing to the relatively stable inpatient mortality rates even as treatment volume grows.
How CAR-T Compares to Stem Cell Transplant
Most patients eligible for CAR-T have already had or been considered for other intensive therapies, including autologous stem cell transplants, where the patient’s own stem cells are harvested and returned after high-dose chemotherapy. A comparison in patients with relapsed or refractory B-cell non-Hodgkin lymphoma found that severe non-blood-related side effects occurred in about 21 percent of CAR-T patients versus 48 percent of those who received transplant.15PubMed Central. Comparison of CAR-T19 and autologous stem cell transplantation for refractory/relapsed non-Hodgkin’s lymphoma That is a meaningful safety advantage, though the two therapies are not always interchangeable and the comparison depends heavily on the specific disease, patient fitness, and timing.
For myeloma, comparison data between salvage transplant and CAR-T have also been generated, with nonrelapse mortality at one year being numerically higher with transplant in some analyses, though the differences have not reached statistical significance given small sample sizes.16Blood. Salvage Transplant Versus CAR-T Cell Therapy for Relapsed Multiple Myeloma The broader point is that CAR-T’s mortality risk, while real, generally compares favorably to the alternatives available to the heavily pretreated patients who receive it.
Differences Between CAR-T Products
Not all CAR-T products carry identical risk profiles. The meta-analysis described earlier found that different products were independently associated with higher nonrelapse mortality for certain diseases. For large B-cell lymphoma, axicabtagene ciloleucel (marketed as Yescarta) was linked to higher NRM, while for multiple myeloma, ciltacabtagene autoleucel (Carvykti) had a similar association.1Nature Medicine. A systematic review and meta-analysis of nonrelapse mortality after CAR T cell therapy Products built with a CD28 costimulatory domain showed numerically higher NRM than those using a 4-1BB domain, at about 8.2 percent versus 6.2 percent, though this difference did not reach conventional statistical significance.17Blood. Infections Drive Non-Relapse Mortality Following CAR-T Therapy across Disease Entities and CAR Products – a Meta-Analysis of Clinical Trials and Real-World Studies
There are also differences based on which protein on cancer cells the therapy targets. A study comparing patients who received BCMA-directed CAR-T (used for myeloma) with those who received CD19-directed CAR-T (used for lymphomas and leukemias) found that BCMA patients had lower rates of certain neurological complications and sepsis but higher rates of liver enzyme elevations. Mortality was comparable between the two groups, at about 4.4 percent for BCMA versus 3.6 percent for CD19.18Clinical Lymphoma, Myeloma and Leukemia. Comparative Clinical Outcomes of BCMA-Directed and CD19-Directed Chimeric Antigen Receptor T-Cell Therapies
Real-World Outcomes Versus Clinical Trials
A common concern with any new therapy is that the carefully selected patients in clinical trials do better than the broader mix of patients treated in routine practice. For CAR-T, the evidence on this has been somewhat reassuring. A systematic review and meta-analysis of real-world data for large B-cell lymphoma found that effectiveness and safety of the two major commercial products were largely consistent with what was seen in their pivotal trials, despite real-world patients being older, sicker, and more diverse.19Transplantation and Cellular Therapy. Real-World Outcomes with Chimeric Antigen Receptor T Cell Therapies in Large B Cell Lymphoma: A Systematic Review and Meta-Analysis That said, phase III trial populations have actually shown numerically higher NRM (about 12 percent) compared to both earlier-phase trials and real-world studies, possibly because phase III designs enrolled patients who might have been excluded from initial studies yet still qualified under broader eligibility criteria.17Blood. Infections Drive Non-Relapse Mortality Following CAR-T Therapy across Disease Entities and CAR Products – a Meta-Analysis of Clinical Trials and Real-World Studies
The Waiting Problem
CAR T-cell therapy is not off the shelf. A patient’s own T cells must be collected, shipped to a manufacturing facility, genetically engineered, expanded, quality-tested, and shipped back. This process takes weeks, sometimes longer. During that wait, patients with aggressive cancers can deteriorate or die before ever receiving their infusion, a reality that does not always show up in the formal mortality statistics of the therapy itself.
A modeling study using data from a pivotal trial of tisagenlecleucel found that reducing wait times by two months could increase the number of eligible patients who actually receive their cells by at least about 11 percent. Among patients who did receive treatment, faster delivery translated into a 3.3 percent improvement in survival gains per treated patient. Combined, the net benefit of a two-month speedup amounted to a roughly 14 percent increase in overall treatment efficacy.20Value in Health. Value of Reducing Wait Times for Chimeric Antigen Receptor T-Cell Treatment: Evidence From Randomized Controlled Trial Data on Tisagenlecleucel for Diffuse Large B-Cell Lymphoma This is a mortality risk that gets overlooked in conversations focused purely on post-infusion side effects, but it represents a real and addressable gap.
The Solid Tumor Challenge
Nearly all of the mortality data discussed so far comes from blood cancers, which is where CAR-T has had its greatest success. Efforts to expand the therapy into solid tumors like lung, breast, or colon cancers face a different and in some ways more dangerous landscape. Solid tumors often share their surface proteins with healthy tissues, meaning the CAR T cells can attack both the tumor and normal organs. This “on-target, off-tumor” toxicity has produced severe and sometimes fatal reactions in early clinical trials targeting solid tumors.21PubMed Central. Overcoming on-target, off-tumour toxicity of CAR T cell therapy for solid tumours
Engineering solutions are under development to address this. Next-generation CAR designs include systems that only fully activate when the T cell encounters two specific signals simultaneously, acting like a molecular “AND gate” that reduces the chance of firing at healthy tissue. Other approaches use inducible switches that let clinicians turn CAR T cells on or off remotely, or inhibitory receptors that tell the T cell to stand down when it encounters a normal-tissue marker.22PubMed Central. Engineering the next generation of CAR T- cells: precision modifications, logic gates and universal strategies to overcome exhaustion and tumor resistance None of these have reached large-scale clinical use yet, but they represent the direction the field is heading as it tries to bring CAR-T’s mortality risk down and extend its reach beyond blood cancers.
Allogeneic CAR-T and “Off-the-Shelf” Risks
Another avenue of active development is allogeneic, or donor-derived, CAR T cells. Instead of engineering each patient’s own cells, these products are manufactured from healthy donor cells and stored for immediate use, eliminating the manufacturing wait. The trade-off is a new category of risk: graft-versus-host disease, where the donor cells attack the recipient’s body. Early clinical data with one allogeneic product showed grade 1 skin graft-versus-host disease in about 10 percent of patients. CRS rates were high, with grade 1 to 2 occurring in 91 percent and more severe forms in 14 percent. Two treatment-related deaths were reported from neutropenic sepsis and pulmonary hemorrhage.23PubMed Central. Addressing graft-versus-host disease in allogeneic cell-based immunotherapy for cancer These are early numbers from small studies, and the field is actively engineering donor cells to reduce graft-versus-host risk through gene editing, but for now, allogeneic products carry their own distinct mortality profile that will need to be sorted out as trials mature.