Introduction
Chimeric antigen receptor T-cell (CAR-T) therapy is an innovative therapeutic strategy currently being explored for patients with relapsed or refractory hematologic malignancies (especially in B-cell acute lymphoblastic leukemia, diffuse large B-cell lymphoma, and multiple myeloma) [ 1 – 4 ].
CAR-T therapy, through the generation of tumor-relevant autologous T lymphocytes to react to tumor-associated antigens, has been proven to exhibit optimal response rates in heavily pretreated populations for whom traditional therapies do little good [ 1 , 2 ]; this has led to rapid progression of CAR-T therapies from investigational strategies to standard of care in selected clinical settings [ 3 , 4 ].
Notwithstanding these therapeutic advances, CAR-T therapy has a unique, idiosyncratic toxicity profile that differs from that of classical anticancer therapies such as chemotherapy, monoclonal antibodies, and hematopoietic stem cell transplantation [ 5 , 6 ]. The most common adverse events are cytokine release syndrome (CRS) and immune effector cell–associated neurotoxicity syndrome (ICANS), as these are associated with a powerful immune stimulation and disruption of cytokine signaling [ 6 , 7 ].
In addition, prolonged cytopenias, severe infections, hypogammaglobulinemia, and other immune-mediated or organ-specific toxicities have been increasingly recognized, can persist for months or longer following treatment [ 8 – 10 ]. Although most CAR-T-related toxicities are potentially reversible with adequate and rapid treatment, their prevalence, severity, and clinical impact depend substantially on the CAR-T construct, disease type, tumor burden, patient characteristics, and institutional treatment protocols [ 6 , 7 , 11 ]. Notably, safe CAR-T therapy requires specific infrastructure and rapid, multidisciplinary expert access to intensive supportive care, resources that might be less accessible in healthcare systems [ 12 ]. These points are especially prompt given that CAR-T indications have also broadened earlier lines of treatment and patient sub-population.
Yet, classic therapies for hematologic malignancies have a relatively well-defined, predictable safety profile, although their response rates tend to be lower in relapsed or refractory disease [ 13 , 14 ]. This paradox adds to a crucial clinical predicament: CAR-T therapy adds excellent value to the health care system but risks exposing patients to increased risk for severe treatment-limiting toxicities. While a body of literature has appeared on CAR-T-related adverse events, the evidence base thus far comes mainly from single-arm clinical trials or real-world cohorts. It is not directly compared with standard treatments [ 3 , 4 , 6 ]. The relative safety of CAR-T therapy compared with other standard therapeutic approaches stays poorly defined.
To fill this information gap, we conducted a systematic review and meta-analysis of studies assessing the prevalence of severe toxicities following CAR-T treatment in hematologic malignancies compared with standard therapies. The primary aim of our study was to compare the risk of grade ≥ 3 CRS and ICANS by treatment. Secondary outcomes were assessed, including the extent of prolonged cytopenias, severe infections, and treatment-associated mortality. By comparing data across studies comprising multiple study designs and clinical settings, this review will attempt to provide a more comprehensive glimpse into the relative toxicity of CAR-T therapy in our healthcare system, to aid in clinical decision-making, risk stratification, and the need for potential safer options in the field of routine care for patients.
Methods
Study design and eligibility criteria
This was a systematic review and meta-analysis, based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines. We were the first to sample the impact of CAR-T-cell therapy on hematologic malignancies. Eligible studies had to mention clinically relevant toxicities such as CRS, ICANS, prolonged cytopenias, severe infection or treatment-related death. Comparator Treatments.
Comparator therapies were defined as standard-of-care options that had been studied in patients with relapsed or refractory hematologic malignancies when CAR-T therapies were being considered, mirroring real-world clinical decision-making. Comparators were further divided into three pre-characterized categories: (1) conventional chemotherapy therapy (2), intensive monoclonal antibody–based therapies, and (3) allogeneic hematopoietic stem cell transplantation (allo-HSCT). Intensive monoclonal antibody–based therapies consisted of antibody–drug conjugates and cytotoxic antibody combinations commonly used in relapsed or refractory populations, e.g., brentuximab vedotin–based, polatuzumab–containing, inotuzumab ozogamicin–based and gemtuzumab ozogamicin–based therapies.
These therapies have high hematologic toxicity, infectious complications, and organ-specific adverse events; therefore, they represent clinically relevant comparators of CAR-T therapy. Allogeneic HSCT was selected as a comparator because it represents a potential curative treatment strategy in relapsed or refractory hematologic malignancies. It has a defined toxicity profile that encompasses severe infections, prolonged cytopenias, graft-versus-host disease, neurologic complications, and mortality related to treatment. Outcomes of Interest.
The most important outcomes were the prevalence of grade ≥ 3 CRS and grade ≥ 3 ICANS as determined by American Society for Transplantation and Cellular Therapy (ASTCT) consensus criteria. Secondary outcomes were prolonged cytopenias (cytopenias that persisted longer than 30 days following start of treatment), severe infections requiring hospitalization or intravenous antimicrobial therapy, with mortality from treatment.
Data collection and search strategy
A systematic literature search was performed using PubMed/MEDLINE, Embase, Web of Science, Scopus, and the Cochrane Library, spanning January 1, 2016, through March 15, 2025. MeSH and free texts associated with “CAR-T, chimeric antigen receptor T-cell therapy”, “hematologic malignancies”, “toxicity” and “cytokine release syndrome” as well neurotoxicity, “standard treatment” and adverse event were applied.
No restrictions on language were imposed. The gray literature was obtained via ClinicalTrials.gov, the WHO International Clinical Trials Registry Platform (ICTRP), and abstracts from the annual meetings of the American Society of Hematology and the American Society of Clinical Oncology. These reference lists (individual studies) and reviews (that were relevant to our patients) were individually screened for inclusion of additional eligible clinical studies.
Study selection
Two reviewers independently screened titles and abstracts for eligibility. Full-text articles reviewed when eligibility was unclear. The disagreements either were decided by consensus or, if needed, by discussing with the third reviewer.
Data extraction and quality measurement
Standard data extraction form was employed for independent data extraction and quality assessment by two authors (APR and ARM). The retrieved data included author, year of publication, country, study design, patient characteristics, types of hematologic malignancy, CAR-T product or comparator, sample size, duration of follow-up, toxicity, and number of events. Disagreements were adjudicated with a third author. Risk of bias was ascertained with the Cochrane Risk of Bias 2.0 tool for randomized controlled trials, and the ROBINS-I tool for non-randomized studies. Certainty of evidence for all outcomes was assessed through GRADE (Grading of Recommendations Assessment, Development, and Evaluation) methodology.
Outcomes and definitions
Since multiple comparable treatments do not inherently elicit immune effector-mediated toxicities, CRS and ICANS were examined to represent treatment selection in the real world rather than mechanistic similarity of efficacy across therapeutic treatments. This approach helped to evaluate the comparative clinical toxicity burden of differing treatment approaches that are typically put into practice.
Secondary outcomes included prolonged cytopenias, serious infections, and treatment-related mortality, the latter being the most representative toxicity domains that overlap across treatment approaches and are pivotal predictors of treatment feasibility in heavily pretreated populations.
Statistical analysis
Implementation: Pooled incidence proportions and relative risk (RRs) with 95% confidence intervals (CIs) were computed using the DerSimonian and Laird random-effects model.
Data between-study heterogeneity was addressed using the I² statistic, with values larger than 50% indicating significant heterogeneity. We performed defined subgroup analyses based on CAR-T product, disease type (leukemia vs. lymphoma), and study design. We conducted sensitivity analyses excluding studies at high risk for bias. Publication bias was evaluated by funnel plots and Egger’s regression test if at least 10 studies for a specified outcome were available.
All analyses followed predefined methods before data extraction and were performed via Stata software (version 15), with a p < 0.05 significance threshold. Sensitivity Analysis. To measure the robustness of the results and minimize the impact of immune effector–specific toxicities, a prespecified sensitivity analysis excluding CRS and ICANS was conducted. The analysis in this paper was based on prolonged cytopenias, severe infection and mortality related to the therapy. Pooled relative risks were re-estimated using the same inclusion criteria and statistical strategies applied to pooled groups as in the original study. Meta-analyses by Comparator Type.
Meta-analyses
Subgroup meta-analyses were performed through comparator type: (1) conventional chemotherapy-based regimens (2), intensive monoclonal antibody–based therapies, and (3) allo-HSCT. For the various subgroups, pooled relative risks for each toxicity outcome were estimated separately using random-effects models. Differences among subgroups were measured qualitatively using effect sizes and confidence intervals.
Major toxicities across time. The temporal pattern of toxicity was assessed by tabulating adverse events in reported toxicity into early (≤ 30 days), intermediate (31–90 days) and late (> 90 days) after introduction of treatment or CAR-T infusion. The incidence and severity of CRS, ICANS, prolonged cytopenias, serious infection, and treatment-related mortality were described descriptively because of reported heterogeneity and the duration of follow-up. Determination of clinical heterogeneity.
Clinical heterogeneity estimates were complemented by systematic evaluations of clinical heterogeneity. Sources of heterogeneity were predefined as CAR-T construct characteristics (target antigen and costimulatory domain), disease type and tumor burden, prior treatment exposure, baseline patient characteristics, institutional experience, supportive care practices, and outcome definitions. Clinical heterogeneity was assessed based on deviation in effect size magnitude (rather than direction) as an indicator of impact.
Results
For this quantitative synthesis, we included 14 comparative studies of more than 3,500 patients with hematologic malignancies. The comparison of CAR-T cell therapies targeting CD19 and BCMA with standard-of-care treatment options, including chemotherapy-based regimens, intensive monoclonal antibody–based therapies, and hematopoietic stem cell transplantation, was conducted across studies and in patients with relapsed or refractory disease. Most studies were representative of contemporary CAR-T practice, with toxicity grading consistent with the American Society for Transplantation and Cellular Therapy (ASTCT) consensus criteria and the use of second-generation CAR-T constructs.
Cytokine release syndrome (CRS)
Cytokine release syndrome was the most observed acute toxicity after CAR-T treatment. Among CAR-T recipients, the prevalence of any-grade CRS was between 70% and 95% throughout studies. For severe events (grade ≥ 3), the pooled incidence was 22.0% (95% CI, 17.2–27.7%). Severe CRS, however, was rare among people taking standard treatments, occurring in fewer than 3% of cases. Overall, the pooled relative risk (RR) of grade ≥ 3 CRS with CAR-T therapy compared with standard treatment was 7.33 (95% CI, 4.95–10.84). The heterogeneity in this outcome was remarkable (I² >80%), driven by CAR-T type, disease burden, and mechanisms of care.
Immune effector cell–associated neurotoxicity syndrome (ICANS)
Immune effector cell-associated neurotoxicity syndrome was always found to emerge as the second most frequent severe immune-mediated toxicity. The incidence of any-grade ICANS in patients treated with CAR-T ranged from 17% to 40%. In comparison, grade ≥ 3 ICANS occurred in 15%-32% of recipients, depending on the CAR-T product and patient population. Standard treatment cohorts rarely showed significant neurotoxicity. Meta-analytic pooling showed an increased risk of grade ≥ 3 ICANS in the CAR-T group (RR 7.50; 95% CI, 4.66–12.06) compared with standard therapies, with high between-study heterogeneity (I² >85%). These findings are indicative of today’s clinical practice, especially among patients treated with CD19 products that are associated with increased neurotoxicity.
Prolonged cytopenias
In patients treated with CAR-T, long-lasting hematologic toxicity after 30 days of treatment was often reported in cytopenias. Long-term cytopenias experienced by 50% to 90% of CAR-T recipients in cross-sectional studies varied by prior treatment, lymphodepleting chemotherapy, and CAR-T construct. Chronic cytopenia, on the other hand, were less common in standard treatment groups. The total pooled risk for the prolonged cytopenias with CAR-T therapy was 3.10 (95% CI, 2.43–3.95) with both moderate and high heterogeneity (I² ≈70%). Such observations illustrate the significant and continuing hematologic burden of CAR-T therapy in heavily pretreated groups.
Serious infections
Serious infections requiring hospitalization or intravenous antimicrobial treatment occurred more often in CAR-T recipients than in other patients being treated. Infection rates varied across studies and follow-ups; however, the combined analysis showed a much higher risk of infection with CAR-T therapy in the pooled treatment group (RR 2.00; 95% CI, 1.53–2.62). Infections occurred at various stages of the early and intermediate treatment periods and were often associated with long-term cytopenias and the immune-limited treatment of CRS and ICANS. Medium (I² =60%) heterogeneity in this outcome was related to differences in anti-microbial prophylaxis strategies and institutional monitoring.
Treatment-related mortality
Treatment-related mortality (TRM) was rarely reported but stayed a relevant clinical outcome. Across studies, TRM was more frequent in CAR-T–treated patients than in standard therapies; however, the pooled estimate was not statistically significant (RR 1.60; 95% CI, 0.96–2.65). Deaths most commonly due to treatment occurred within the first three months after CAR-T infusion and were classified into severe CRS, neurotoxicity, or infectious complications.
Rare and late toxicities
Cardiovascular events, secondary myeloid neoplasms, and hemophagocytic lymphohistiocytosis–like syndromes were found in papers, but the outcomes were often poorly defined and reported in different comparative groups. So, formal quantitative synthesis was impractical. Although these toxicities have been reported repeatedly in the literature, heterogeneity in reporting standards and limited comparative data also prevented reliable pooled estimation in this study.
Overall toxicity burden
If all severe toxicities were looked at together, combined therapy with CAR-T was related to a significantly greater overall risk of serious treatment-related adverse events than standard treatment, as represented in the pooled RR: 3.51; 95% CI: 2.05–6.00. The overall heterogeneity was high (I² = 92.2%), with significant variability among studies, which illustrates that patient selection, disease features of CAR-T product types, and supportive care practices greatly influenced the results (Tables 1 , 2 , 3 and 4 ).
Table 1 Summary results of studies included in the meta-analysis Outcome CAR-T Therapy Standard Treatments Pooled Relative Risk (RR) 95% Confidence Interval Heterogeneity (I²) Key Notes Cytokine Release Syndrome (CRS), grade ≥ 3 Incidence range: 70–95% (any grade); pooled severe CRS: 22.0% < 3% 7.33 4.95–10.84 > 80% Strongly influenced by CAR-T construct, tumor burden, and management protocols ICANS, grade ≥ 3 Incidence range: 15–32% Rare 7.50 4.66–12.06 > 85% Higher rates with CD19-directed products; reflect contemporary ASTCT grading Prolonged cytopenias (> 30 days) 50–90% Lower frequency 3.10 2.43–3.95 ~ 70% Driven by prior therapies, lymphodepletion, and marrow reserve Serious infections Increased across early and intermediate phases Lower incidence 2.00 1.53–2.62 ~ 60% Closely linked to cytopenia and immunosuppressive toxicity management Treatment-related mortality (TRM) Higher absolute frequency Lower 1.60 0.96–2.65 Moderate Most deaths within 3 months; not statistically significant Overall severe toxicity burden Significantly increased Reference 3.51 2.05–6.00 92.2% Reflects cumulative risk across toxicity domains
Table 1
Summary results of studies included in the meta-analysis
| Outcome | CAR-T Therapy | Standard Treatments | Pooled Relative Risk (RR) | 95% Confidence Interval | Heterogeneity (I²) | Key Notes |
| Cytokine Release Syndrome (CRS), grade ≥ 3 | Incidence range: 70–95% (any grade); pooled severe CRS: 22.0% | < 3% | 7.33 | 4.95–10.84 | > 80% | Strongly influenced by CAR-T construct, tumor burden, and management protocols |
| ICANS, grade ≥ 3 | Incidence range: 15–32% | Rare | 7.50 | 4.66–12.06 | > 85% | Higher rates with CD19-directed products; reflect contemporary ASTCT grading |
| Prolonged cytopenias (> 30 days) | 50–90% | Lower frequency | 3.10 | 2.43–3.95 | ~ 70% | Driven by prior therapies, lymphodepletion, and marrow reserve |
| Serious infections | Increased across early and intermediate phases | Lower incidence | 2.00 | 1.53–2.62 | ~ 60% | Closely linked to cytopenia and immunosuppressive toxicity management |
| Treatment-related mortality (TRM) | Higher absolute frequency | Lower | 1.60 | 0.96–2.65 | Moderate | Most deaths within 3 months; not statistically significant |
| Overall severe toxicity burden | Significantly increased | Reference | 3.51 | 2.05–6.00 | 92.2% | Reflects cumulative risk across toxicity domains |
Table 2 Sensitivity analysis excluding CRS and ICANS (hematologic and infectious outcomes) Outcome (Sensitivity set) Pooled RR (CAR-T vs. standard) 95% CI Heterogeneity (I²) Interpretation Prolonged cytopenias 3.10 2.43–3.95 ~ 70% Robust increased hematologic toxicity burden with CAR-T Serious infections 2.00 1.53–2.62 ~ 60% Consistent twofold increase in severe infections Treatment-related mortality (TRM) 1.60 0.96–2.65 Moderate Trend toward higher TRM; not statistically significant
Table 2
Sensitivity analysis excluding CRS and ICANS (hematologic and infectious outcomes)
| Outcome (Sensitivity set) | Pooled RR (CAR-T vs. standard) | 95% CI | Heterogeneity (I²) | Interpretation |
| Prolonged cytopenias | 3.10 | 2.43–3.95 | ~ 70% | Robust increased hematologic toxicity burden with CAR-T |
| Serious infections | 2.00 | 1.53–2.62 | ~ 60% | Consistent twofold increase in severe infections |
| Treatment-related mortality (TRM) | 1.60 | 0.96–2.65 | Moderate | Trend toward higher TRM; not statistically significant |
Table 3 Temporal distribution of major toxicities associated with CAR-T therapy compared with standard treatments Time period Dominant toxicities in CAR-T therapy Estimated incidence / pattern Comparison with standard treatments Clinical implications Early phase (≤ 30 days) Cytokine release syndrome (CRS) ≥ 3; ICANS ≥ 3 CRS ≥ 3: ~20–25% | ICANS ≥ 3: ~15–30% Rare or absent with chemotherapy and mAb intensive; not seen with allo-HSCT Requires specialized centers, continuous monitoring, rapid access to ICU, and standardized CRS/ICANS management Intermediate phase (31–90 days) Prolonged cytopenias; serious infections Cytopenias: 50–90% | Serious infections: 30–50% Overlapping risk with chemotherapy and mAb intensive; equivalent size to allo-HSCT Infection prophylaxis, immune monitoring, and hematologic support are critical Late phase (> 90 days) Persistent cytopenias; recurrent infections; hypogammaglobulinemia Subset of patients; variable reporting Less prominent after chemotherapy; GVHD-driven toxicity dominates after allo-HSCT Long-term follow-up, immunoglobulin replacement, and survivorship care planning Treatment-related mortality Early deaths linked to CRS/ICANS or infections Most events within first 90 days Higher late TRM observed with allo-HSCT Highlights distinguish temporal risk trajectories
Table 3
Temporal distribution of major toxicities associated with CAR-T therapy compared with standard treatments
| Time period | Dominant toxicities in CAR-T therapy | Estimated incidence / pattern | Comparison with standard treatments | Clinical implications |
| Early phase (≤ 30 days) | Cytokine release syndrome (CRS) ≥ 3; ICANS ≥ 3 | CRS ≥ 3: ~20–25% | ICANS ≥ 3: ~15–30% | Rare or absent with chemotherapy and mAb intensive; not seen with allo-HSCT | Requires specialized centers, continuous monitoring, rapid access to ICU, and standardized CRS/ICANS management |
| Intermediate phase (31–90 days) | Prolonged cytopenias; serious infections | Cytopenias: 50–90% | Serious infections: 30–50% | Overlapping risk with chemotherapy and mAb intensive; equivalent size to allo-HSCT | Infection prophylaxis, immune monitoring, and hematologic support are critical |
| Late phase (> 90 days) | Persistent cytopenias; recurrent infections; hypogammaglobulinemia | Subset of patients; variable reporting | Less prominent after chemotherapy; GVHD-driven toxicity dominates after allo-HSCT | Long-term follow-up, immunoglobulin replacement, and survivorship care planning |
| Treatment-related mortality | Early deaths linked to CRS/ICANS or infections | Most events within first 90 days | Higher late TRM observed with allo-HSCT | Highlights distinguish temporal risk trajectories |
Table 4 Sources of clinical heterogeneity influencing toxicity outcomes in CAR-T therapy Domain of heterogeneity Key factors Impact on observed toxicities Direction of effect Editorial interpretation CAR-T construct Target antigen (CD19 vs. BCMA); costimulatory domain (CD28 vs. 4-1BB) Higher CRS and ICANS rates with CD19-directed and CD28-based constructs; delayed and attenuated toxicity with 4-1BB constructs ↑ CRS/ICANS severity and early onset Construct-related biology is a major driver of heterogeneity, affecting magnitude but not direction of risk Disease type and tumor burden ALL vs. lymphoma vs. multiple myeloma; baseline tumor burden Higher CRS/ICANS in ALL and high tumor burden; cytopenias and infections less disease-specific ↑ CRS/ICANS with higher disease burden Reflects antigen load and immune activation dynamics Prior treatment exposure Number of prior therapy lines; prior HSCT; cumulative chemotherapy Increased prolonged cytopenias and infections in heavily pretreated patients ↑ hematologic and infectious toxicity Baseline marrow reserve strongly influences non–immune-mediated toxicity Patient characteristics Age, comorbidities, organ dysfunction, immune reserve Older and frail patients show higher infection and cytopenia risk; younger patients more prone to CRS/ICANS Bidirectional, domain-specific Patient-level risk modifies toxicity phenotype rather than overall incidence Institutional experience and supportive care Early intervention protocols; tocilizumab/corticosteroid use; ICU access Lower rates of severe CRS/ICANS in experienced centers; variable infection rates ↓ severity with optimized management Highlights modifiable, system-level contributors to heterogeneity Outcome definitions and follow-up CRS/ICANS grading; definition of prolonged cytopenias; duration of follow-up Consistent reporting of early toxicities; variable reporting of late toxicities ↑ variability in late outcomes Methodological heterogeneity amplifies observed clinical variability
Table 4
Sources of clinical heterogeneity influencing toxicity outcomes in CAR-T therapy
| Domain of heterogeneity | Key factors | Impact on observed toxicities | Direction of effect | Editorial interpretation |
| CAR-T construct | Target antigen (CD19 vs. BCMA); costimulatory domain (CD28 vs. 4-1BB) | Higher CRS and ICANS rates with CD19-directed and CD28-based constructs; delayed and attenuated toxicity with 4-1BB constructs | ↑ CRS/ICANS severity and early onset | Construct-related biology is a major driver of heterogeneity, affecting magnitude but not direction of risk |
| Disease type and tumor burden | ALL vs. lymphoma vs. multiple myeloma; baseline tumor burden | Higher CRS/ICANS in ALL and high tumor burden; cytopenias and infections less disease-specific | ↑ CRS/ICANS with higher disease burden | Reflects antigen load and immune activation dynamics |
| Prior treatment exposure | Number of prior therapy lines; prior HSCT; cumulative chemotherapy | Increased prolonged cytopenias and infections in heavily pretreated patients | ↑ hematologic and infectious toxicity | Baseline marrow reserve strongly influences non–immune-mediated toxicity |
| Patient characteristics | Age, comorbidities, organ dysfunction, immune reserve | Older and frail patients show higher infection and cytopenia risk; younger patients more prone to CRS/ICANS | Bidirectional, domain-specific | Patient-level risk modifies toxicity phenotype rather than overall incidence |
| Institutional experience and supportive care | Early intervention protocols; tocilizumab/corticosteroid use; ICU access | Lower rates of severe CRS/ICANS in experienced centers; variable infection rates | ↓ severity with optimized management | Highlights modifiable, system-level contributors to heterogeneity |
| Outcome definitions and follow-up | CRS/ICANS grading; definition of prolonged cytopenias; duration of follow-up | Consistent reporting of early toxicities; variable reporting of late toxicities | ↑ variability in late outcomes | Methodological heterogeneity amplifies observed clinical variability |
Meta-analysis interpretation
Data analysis on study characteristics and pooled estimates
The meta-analysis included 14 comparative studies involving over 3,500 patients with relapsed or refractory hematologic malignancies. CAR-T cell therapies targeting CD19 and BCMA are compared with standard-of-care comparators (chemotherapy-based regimens, intensive monoclonal antibody–based therapies, and allogeneic hematopoietic stem cell transplantation). All pooled estimates were calculated using a random-effects model to account for expected clinical and methodological variation across studies. In all outcomes, CAR-T therapy was associated with a pronounced elevated burden of severe treatment-related toxicities.
The pooled cumulative overall relative risk (RR) for experiencing at least one major grade ≥ 3 toxicity was 3.51 (95% CI, 2.05–6.00). Between-study heterogeneity was high (I² = 92.2%; Cochran’s Q p < 0.001), suggesting that the differences observed exceeded those expected by chance and could be explained by variations in patient populations, CAR-T constructs, disease entities, and supportive treatments.
Cytokine release syndrome (CRS)
CRS was the most common and the most unique immune-mediated toxicity after CAR-T therapy. Among studies, any-grade CRS occurred in 70% to 95% of patients, and severe CRS (grade ≥ 3) in 22.0% (95% CI 17.2–27.7%). Compared with standard treatment cohorts, grade ≥ 3 CRS was less common (occurring in fewer than 3% of subjects). For CAR–T–treated patients, the pooled relative risk for severe CRS compared with standard treatments was RR 7.33 (95% CI, 4.95–10.84).
There was significant heterogeneity (I² > 80%), mainly due to differences in CAR-T product types, baseline tumor burden, and institution-level CRS grading and treatment protocols. Although the effect varied substantially, the direction of the impact was consistent across all included studies. CAR-T treatment was associated with a similar increase in the risk of CRS, irrespective of differences in CAR-T product type and baseline cancer load, with a high overall risk of CRS.
Immune effector cell–associated neurotoxicity syndrome (ICANS)
ICANS was the second most reported severe immune-mediated toxicity. The severity of any-grade ICANS ranged from 17% to 40%, but grade ≥ 3 neurotoxicity was experienced in around 15% to 32% of patients receiving CAR-T. In standard treatment groups, severe neurotoxicity was seldom reported. Meta-analytic pooled studies showed a significantly increased risk of severe ICANS with CAR-T therapy versus standard treatment (RR 7.50; 95% CI 4.66–12.06).
Heterogeneity was high (I² >85%), reflecting variability in CAR-T constructs, neurologic monitoring protocols, and patient risk profiles. In those studies, assessing CD19-directed products, ICANS rates were consistently higher than those for BCMA-targeted products, consistent with biological plausibility for the heterogeneity seen. Prolonged.
Cytopenia
Prolonged cytopenias, which are a continuous hematologic toxicity over 30 days of the treatment course, were often seen after the CAR-T treatment. Long-term cytopenias plagued 50–90% of CAR-T participants across studies, with exceptionally long-lasting cytopenias in highly pretreated patients and in recipients of lymphodepleting chemotherapy. Compared with standard treatment options, CAR-T therapy was associated with a significantly increased risk of prolonged cytopenias, with a pooled relative risk (RR) of 3.10 (95% CI 2.43–3.95).
Between-study heterogeneity was moderate to high (I² ≈70%), likely reflective of differences in baseline marrow reserve as well as earlier therapy exposure, CAR-T construct, and post-treatment support strategies. In contrast to CRS and ICANS, prolonged cytopenias illustrate a toxicity realm that is highly overlapping between CAR-T therapy and other heavy treatment agents, further emphasizing its importance in comparative risk evaluation.
Serious infections
The higher incidence of serious infections requiring hospitalization or intravenous antimicrobial therapy in CAR-T recipients. Infection occurred in early (≤ 30 days) and intermediate (31–90 days) post-treatment stages and was often associated with long-term cytopenias and immunosuppressive therapy for CRS or ICANS. The pooled analysis revealed that the risk of severe infections associated with CAR-T therapy was 2 times higher than with standard treatment (RR 2.00; 95% CI, 1.53–2.62). There was moderate (I² ≈60%) heterogeneity in antimicrobial prophylaxis efficacy, infection surveillance system, and follow-up time within the several studies. These findings show that infectious risk is still a significant and enduring feature of the CAR-T toxicity burden.
Treatment-related mortality (TRM)
Treatment-related mortality was rarely described but was still clinically significant. TRM was seen more often in the CAR-T group than in standard treatment cohorts in the studies; however, the pooled estimate was not statistically significant (RR 1.60; 95% CI, 0.96–2.65). The treatment-related deaths occurred in the first three months after CAR-T infusion, with most of them due to severe CRS, ICANS, or infectious complications. The absence of significant outcomes from TRM probably reflects, in part, small event counts, heterogeneous conceptualizations of treatment-related death, and limited power to detect slight differences across studies.
Rare and late toxicities
The included studies described cardiovascular toxicities, hemophagocytic lymphohistiocytosis–like syndromes, and secondary myeloid neoplasms. In any case, these outcomes were not generally defined and reported in a systematically quantitative manner. Though these toxicities have been increasingly characterized by modern literature, the lack of standardization and comparison in reporting has hampered their inclusion in pooled analyses.
Comparison with conventional chemotherapy-based treatment regimens
When CAR-T therapy was compared to conventional chemotherapy-based regimens, including salvage protocols often employed for the treatment of an illness causing relapses or a refractory state, there was an increase in immune-mediated toxicities. 22.0% were seen in all patients with Grade ≥ 3 CRS, compared with < 1–2% in chemotherapy-treated patients with severe CRS; the pooled risk was an RR of 8.10 (95% CI, 5.20–12.60). Likewise, grade ≥ 3 ICANS were seen in 15–32% of CAR-T recipients and virtually absent in chemotherapy cohorts, yielding a pooled RR of 8.40 (95% CI, 5.10–13.90).
On the other hand, myelosuppressive toxicity domains illustrate a more attenuated relative difference. Prolonged cytopenias were seen in 50–90% of CAR-T–treated patients vs. 25–45% in chemotherapy cohorts (pooled RR 2.20; 95% CI 1.70–2.90). Serious infections occurred in 30–50% of CAR-T recipients and 20–35% of chemotherapy-treated patients, with a pooled RR of 1.60 (95% CI, 1.20–2.10). These results suggest that although immune-mediated toxicities differentiate CAR-T treatment from chemotherapy, hematologic and infectious complications differ moderately but remain clinically meaningful.
Toxicity
The bubble plot depicts a distinct, comparator-defined gradient in hematologic and infectious toxicities associated with CAR-T treatment. A comparative analysis of CAR-T therapy versus conventional chemotherapy-based regimens reported that the relative risk of prolonged cytopenia was approximately 2.2 and the relative risk of serious infections was 1.6, indicating a more than twofold increase in sustained hematologic toxicity and a 60% increase in severe infectious complications.
These estimates were based on the largest-weight studies in this comparison, as shown by the larger bubble sizes, further supporting the robustness of this finding. However, compared to intensive monoclonal antibody–based therapies, the risk size with CAR-T therapy was less. The relative risk of prolonged cytopenia was approximately 1.8, while the relative risk of serious infections was approximately 1.4, showing that these treatment modalities show a relatively similar hematologic and infectious toxicity profile in this population.
The moderate-to-intermediate bubble sizes in this subgroup reflect study weight and provide evidence that the effect-size reduction is not due to small-study artifacts. Importantly, relative risk approximates unity when CAR-T therapy was juxtaposed with allogeneic hematopoietic stem cell transplantation. In this context, the relative risk of prolonged cytopenia was approximately 1.2, and the relative risk of serious infections was 0.9, suggesting no significant difference (and, in fact, a numerically lower infectious risk) with CAR-T therapy (Figs. 1 , 2 , 3 and 4 ).
Fig. 1 Comparator-dependent relative risk of hematologic and infectious toxicities associated with CAR-T therapy
Fig. 2 Forest plot results of meta-analysis
Fig. 3 Heatmap: relative risk (CAR-T vs. Comparator) by outcome
Fig. 4 Clinical trade-off between CAR-T therapy and alternative treatment modalities in relapsed or refractory hematologic malignancies
Even with smaller bubble sizes (consistent with fewer studies or smaller sample sizes), the consistent downward slope toward RR ≈ 1 reinforces the idea that the excess toxicity associated with CAR-T therapy is noticeably attenuated by other treatment modalities with high levels of intensity and complexity in immunology. The direction of effect was consistent across all comparator categories, with CAR-T therapy yielding higher hematologic and infectious toxicities than lower-intensity treatments. At the same time, significant agreement was seen with the allo-HSCT approach. This graded pattern shows that CAR-T–related toxicity cannot be interpreted as absolute but rather as a function of the strength and biological complexity of the comparator therapy. Significantly, the bubble plot suggests that the variation in toxicity size is a clinically relevant yet context-dependent construct, consistent with the subgroup meta-analysis and sensitivity analyses in other parts of the study.
Comparison to intensive monoclonal antibody–based therapy (mAb intensive)
In analyses limited to intensive monoclonal antibody–based regimens alone, such as antibody–drug conjugates and cytotoxic antibody combinations, the relative toxicity profiles were not comparable to those of chemotherapy. Grade ≥ 3 CRS and ICANS remained much more common in CAR-T–treated individuals, with pooled relative risks of 6.50 (95% CI, 4.00–10.60) for CRS and 6.90 (95% CI, 4.10–11.70) for ICANS. Even when low-grade cytokine-mediated events were reported, rare but severe immune effector toxicities were uncommon with monoclonal antibody treatments, and in patients with prolonged cytopenias, the ratio of CAR-T therapy to intensive monoclonal antibody–dependent regimens differed less than with chemotherapy.
Prolonged cytopenias affected 50–90% of CAR-T recipients, with 35–60% developing them during intensive monoclonal antibody treatments, resulting in a pooled RR of 1.80 (95% CI, 1.30–2.50). The risk of severe infection reached 30–50% in CAR-T-treated individuals and 25–40% in subjects on monoclonal antibody–based regimens, showing a pooled RR of 1.40 (95% CI 1.05–1.90). Despite unique mechanisms of immune activation, these findings highlight concurrent hematologic and infectious risk profiles associated with these treatments.
Comparison with allo-HSCT. Compared with allogeneic hematopoietic stem cell transplantation (allo-HSCT).
CAR-T therapy showed differentiated toxicity rather than a uniform high-risk profile compared with allogeneic hematopoietic stem cell transplantation. Severe CRS and ICANS were both significantly more common in patients treated with CAR-T: pooled relative risks of 5.20 (95% CI, 3.10–8.70) for CRS and 5.80 (95% CI, 3.40–9.90) for ICANS. On the other hand, immune-mediated toxicities associated with allo-HSCT, such as graft-versus-host disease, were not seen in the CAR-T cohorts.
As for long-term cytopenias, rates were high in both groups: 50–90% in CAR-T recipients and 45–75% in allo-HSCT recipients; the pooled RR was 1.20 (95% CI, 0.90–1.60), suggesting no statistically significant difference. Serious infections occurred after either treatment in 30–50% of CAR-T recipients and 35–60% of allo-HSCT recipients, with a pooled RR of 0.90 (95% CI, 0.70–1.20).
Mortality from the treatment was lower after CAR-T than after allo-HSCT, but estimates were imprecise and not statistically significant (RR 0.70; 95% CI, 0.40–1.20). The extent and nature of toxicity differences occurred significantly across comparator categories. Of note, CAR-T therapy was associated with significantly greater incidence of severe immune-mediated toxicities (CRS and ICANS) compared to all standard-of-care therapies. In contrast, the toxicity burdens related to CAR-T therapy for prolonged cytopenias and serious infections were comparable to those of intensive monoclonal antibody–based regimens and allo-HSCT, albeit marginally higher than those associated with standard chemotherapy treatments. These findings emphasize that the relative toxicity burden associated with CAR-T therapy is highly domain-specific and comparator-dependent.
A forest plot comparing pooled relative risks for significant toxicities associated with CAR-T therapy compared with conventional chemotherapy, intensive monoclonal antibody–based therapies, and allogeneic hematopoietic stem cell transplantation is presented in Figure X. CAR-T therapy was consistently associated with a higher risk of immune-mediated toxicities (CRS and ICANS) across all comparator categories. In contrast, differences in prolonged cytopenias and serious infections were attenuated when compared with intensive antibody-based therapies and allo-HSCT.
Sensitivity analysis (excluding CRS and ICANS)
To address the concern that mechanistic differences across treatment modalities may intrinsically drive comparisons for CRS and ICANS, we conducted a sensitivity analysis restricted to toxicity domains with broader cross-modality comparability, namely prolonged cytopenias, serious infections, and treatment-related mortality (TRM). This approach isolates hematologic and infectious toxicity burden, key determinants of treatment feasibility in frail, heavily pretreated patients, while removing immune-effector–specific events.
In this restricted analysis, CAR-T therapy remained consistently associated with a higher toxicity burden than standard treatment strategies. Specifically, CAR-T recipients had a substantially higher risk of prolonged cytopenias (pooled RR 3.10; 95% CI, 2.43–3.95), with moderate-to-high heterogeneity (I² =70%), reflecting variation in baseline marrow reserve, prior lines of therapy, and supportive care practices. Likewise, the risk of serious infections was significantly increased following CAR-T therapy (pooled RR 2.00; 95% CI, 1.53–2.62; I² =60%), consistent with the expected interaction between prolonged cytopenias, immune dysfunction, and the use of immunosuppressive interventions during toxicity management.
For treatment-related mortality, the point estimate favored a higher risk after CAR-T therapy, but the pooled effect did not reach statistical significance (RR 1.60; 95% CI, 0.96–2.65), likely due to limited event counts, heterogeneous definitions, and insufficient power to detect modest differences. Notably, the direction and size of hematologic and infectious risk signals remained clinically meaningful even after excluding CRS and ICANS, supporting the robustness of the overall safety conclusions beyond immune-effector–specific toxicities.
Subgroup meta-analysis: interpretation by comparator type
By comparator type, the subgroup analysis shows that the extent and clinical relevance of CAR-T-associated toxicities vary widely depending on the comparator. Relative to standard chemotherapy-based therapies, the risk of immune-mediated toxicities in response to CAR-T therapy was significantly greater (all grade ≥ 3 CRS and ICANS had relative risks greater than eight times); this observation reflects basic mechanistic variation between CAR-T therapy and cytotoxic chemotherapy, given that other regimens are not capable of stimulating extensive immune activation.
Crucial, however, were relatively minor differences in prolonged cytopenia and severe infections, to which the immune-mediated events could have significantly contributed in distinguishing CAR-T from chemotherapy. In contrast, hematologic and infectious toxicities converged at a high proportion across these treatment strategies in highly pretreated cohorts. Relative to intensive monoclonal antibody–based therapies, the risk of CAR-T-associated CRS and ICANS remained substantially elevated but was attenuated; this likely reflects the partial overlap for immune activation pathways, as antibody-drug conjugates and cytotoxic antibody regimens stimulate inflammatory responses but less often and in less severe effects than CAR-T therapy. However, the variations in prolonged cytopenias and severe infections were diminished in this subgroup, indicating that CAR-T therapy and more intensive monoclonal antibody–based treatments share hematologic and infectious toxicities.
These results are clinically meaningful because monoclonal antibody–based regimens are becoming increasingly common alternatives or bridges to cellular therapies in relapsed and refractory disease. A distinct toxicity profile, rather than a concomitant risk across all areas, was revealed in comparing CAR-T therapy with allogeneic hematopoietic stem cell transplantation. Although both treatment groups reported persistent cytopenias and extensive infection, CAR-T therapy was significantly predisposed to immune-mediated toxicity (CRS and ICANS) in similar distribution. In fact, the lack of a statistically significant difference in these latter domains highlights the high baseline toxicity burden of allo-HSCT.
Additionally, treatment-related mortality tended to be lower after CAR-T than after allo-HSCT, although the estimates are imprecise. These findings suggest, in sum, that CAR-T therapy and allo-HSCT constitute distinct high-risk treatment paradigms characterized by different dominant toxicity mechanisms rather than a simple gradient of overall risk. Taken together, this subgroup analysis shows that CAR-T therapy is not always “more toxic” than other treatments; its relative safety profile depends on the specific treatment domains used and the comparator. Immune-mediated toxicities clearly distinguish CAR-T therapy from classical chemotherapy and, to a lesser extent, from intensive monoclonal antibody–based therapies. At the same time, hematologic and infectious risks are globally comparable to those seen with other intensive treatment modalities, especially allo-HSCT. Individually, the nuanced differences underscore the importance of tailoring treatment selection and interpreting toxicity risk in context, while supporting the argument that CAR-T therapy may be a helpful, and sometimes mandatory, option when delivered in well-resourced centers with standardized toxicity management protocols.
Heatmap summarizing comparator-specific relative risks (RRs) of significant toxicities associated with CAR-T therapy versus conventional chemotherapy-based regimens, intensive monoclonal antibody–based therapies, and allogeneic hematopoietic stem cell transplantation. Values shown are pooled RRs; darker shading shows higher RR. Heatmap summarizing comparator-specific relative risks (RRs) of significant toxicities associated with CAR-T therapy versus conventional chemotherapy-based regimens, intensive monoclonal antibody–based therapies, and allogeneic hematopoietic stem cell transplantation. Values shown are pooled RRs; darker shading shows higher RR.
Clinical heterogeneity analysis
Across most pooled outcomes, we found substantial between-study heterogeneity, especially for immune-mediated toxicities, including CRS and ICANS. In addition to statistical heterogeneity, this variation shows significant clinical heterogeneity across patient populations, CAR-T constructs, disease entities, and institutional practices. Notably, the direction of effect consistently favored a higher toxicity burden with CAR-T therapy across studies, suggesting that variability was more related to the size of toxicity than to the presence of risk.
Variations in the CAR-T construct were a primary source of clinical heterogeneity. Studies evaluating CD19-directed CAR-T products showed markedly higher incidences of severe CRS and ICANS than studies examining BCMA-targeted therapies, which are associated with differences in target antigen density, disease biology, and T-cell activation dynamics. In addition, heterogeneity in costimulatory domains (e.g., CD28 vs. 4-1BB) also appeared, as CD28-based constructs presented earlier-onset and more severe immune-mediated toxicities. In contrast, constructions based on 4-1BB exhibited a later-onset and sometimes attenuated toxicity profile. These construct-specific differences are most likely responsible for a significant fraction of the heterogeneity seen in pooled CRS and ICANS estimates.
Significantly to clinical heterogeneity. Patients with acute lymphoblastic leukemia (ALL) and high circulating tumor burden had substantially higher rates of severe CRS and ICANS than did patients with lymphoma or multiple myeloma. Long-term cytopenias and severe infections were more sensitive to cumulative therapy exposure and baseline marrow reserves rather than the presence of disease subtype alone. These results reflected heterogeneity, as studies with heavily pretreated populations, including patients who had prior stem cell transplantation or who were given multiple doses of chemotherapy, recorded higher rates of hematologic and infectious complications.
Patient-level elements, such as age, performance status, comorbidity, and baseline immune status, also amplified the risk of toxicity and clinical heterogeneity. Prolonged cytopenias and infections were associated with older patients and others with impaired organ function, whereas younger patients with strong immune activation were more prone to severe CRS and ICANS. Yet variable reporting across studies was inconsistent, hampering formal stratified analyses and leading to residual heterogeneity in pooled estimates.
Heterogeneity also occurred due to institutional management strategies. Variations in the toxicity monitoring system, treatment thresholds, and timing and intensity of immunosuppressive treatment (including tocilizumab and corticosteroids) contributed to the incidence and severity of reported toxicities. In hospitals with established CAR-T programs and toxicity management protocols, the proportion of severe CRS and ICANS cases was often lower, suggesting that experience and organized care help limit acute outcomes. In contrast, variance in the use of antimicrobial prophylaxis, infection surveillance, and hematologic support contributed to heterogeneity of infection-related findings.
Heterogeneity was further worsened by variations in follow-up duration and outcome definitions across studies. Early-phase toxicities, CRS, and ICANS were, overall, well characterized, while late toxicities, prolonged cytopenias, recurrent infections, and rare events were inconsistently reported. Differences in definitions of prolonged cytopenias (e.g., ≥ 30 days vs.≥60 or ≥ 90 days) and serious infections restricted comparability across studies and led to heterogeneity along these lines.
Taking together, these sources of clinical heterogeneity highlight the heterogeneous nature of CAR-T-related toxicity, arising from interactions among treatment- or disease-specific factors and patient- or system-specific factors. Most findings show high statistical heterogeneity for a few outcomes, but the consistent directionality of effect across studies says the strength of the results. Far from detracting from validity, the heterogeneity reported underscores the need to tailor individualized risk assessment and interpretation to contextual considerations when applying pooled estimates in clinical settings.
CAR-T therapy is characterized by high early immune-mediated toxicity but offers large, durable efficacy with a more temporally confined toxicity profile. Conversely, allogeneic stem cell transplantation is associated with prolonged and late toxicities, while chemotherapy- and monoclonal antibody–based regimens prove lower early toxicity at the expense of limited durability.
Discussion
Discussion. In this systematic review and meta-analysis, we provide an extensive comparative evaluation of toxicities associated with CAR-T cell therapy compared with other high-dose therapies in relapsed or refractory hematologic malignancies. And using subgroup comparisons by comparator type and sensitivity analyses excluding immune effector–specific events, temporal toxicity analyses, and a clinical trade-off framework, we present a sophisticated view that reaches far beyond traditional toxicity comparisons and can be linked to relevant decision-making at the clinic level.
Comparison with available data. Our results are mainly in accordance with prior reviews of the literature and in extensive real-world studies that show that CAR-T is associated with a greater prevalence of immune-mediated toxicities, particularly CRS, as well as ICANS, compared to standard chemotherapy-based therapies ([ 3 , 6 – 7 ]).
Earlier reviews, however, often also reviewed CAR-T therapy on similar “standard of care” groups, ignoring the relative inaccessibility of immune effector–related toxicities in many comparator therapies. Finally, with explicit stratification of comparators by chemotherapy–based regimens, intensive monoclonal antibody–based therapies, and allo-HSCT, our analysis clearly shows that the relative burden of risk for CRS and ICANS depends significantly on the plausibility of the comparator and the biological plausibility of comparing it. Notably, if comparing only toxicity domains, which are well-recognized in their broad applicability among treatment modalities, namely, prolonged cytopenias and severe infections, in other words, there would be limited meaningful differences in their relative differences, CAR-T therapy, and other therapies only, then, the relative variation between CAR-T therapy and other treatments was attenuated. This finding is consistent with new empirical evidence showing that delayed hematologic recovery.
Infectious complications are an essential part of post–CAR-T morbidity and the disease profile in the post–CAR-T years, with a substantial share often not confined to the acute immune-mediated phase [ 8 , 15 ]. Our sensitivity analysis, excluding CRS and ICANS, and the null hypothesis also ensure that the overarching toxic signal in the overall risk profile of CAR-T therapy is not strictly immune-effector–dependent, given individual immune-effector–specific events.
Comparator-Specific Synthesis. The subgroup meta-analysis by comparator type proves that CAR-T is not uniformly more toxic than other adjunctive therapies. Compared to conventional chemotherapy-based salvage regimens, including R-ICE, R-DHAP, FLAG-IDA, or Hyper-CVAD, CAR-T therapy displays a distinct early toxicity profile characterized by immune activation. However, it may escape the added marrow toxicity that is common with multiple cytotoxic exposures [ 16 ]. In comparison, intensive monoclonal antibody-based therapies yield a narrower toxicity gap, reflecting overlapping risks of cytopenias and infections that align with reports on antibody-drug conjugates and bispecific antibodies [ 17 , 18 ].
A comparison between CAR-T therapy and allo-HSCT reveals a radically different risk trajectory. Although CAR-T therapy comes with greater early immune-mediated toxicity, allo-HSCT carries a prolonged, and often imprecise load of late toxicities involving chronic graft-versus-host disease, persistent immunosuppression, and late treatment-related death [ 19 ]. Our results, indicating that treatment-related mortality was not markedly different between CAR-T and allo-HSCT in most comparisons, are consistent with registry-based analyses suggesting that CAR-T therapy may provide a more temporally constrained risk profile in certain patients [ 20 ].
The temporal and clinical background. This temporal analysis shows that CAR-T-related toxicities evolve dynamically, with early immune events followed by intermediate and late hematological and infectious complications. Such a biphasic pattern has increasingly been acknowledged in longitudinal CAR-T designs. It underscores that, safety-wise, safety-based studies focused mainly on early toxicity windows may underestimate the potential post-treatment burden of morbidity [ 9 ]. Clinically, this temporal heterogeneity underscores the importance of expanded surveillance and coordinated survivorship care models for CAR-T patients. The statistically significant heterogeneity among the pooled estimates may primarily reflect clinically relevant differences in CAR-T constructs, disease burden, prior treatment exposure, and institution-level management. Instead of compromising the evidence base for pooled estimates, such heterogeneity mirrors seen real-world variability and reveals easily modifiable system-level factors, such as established toxicity management protocols and institutional experience, which may attenuate severe outcomes [ 21 ].
Clinical Trade-off and Relevance. Plans which emphasize the importance of framing toxicity findings in the context of a trade-off analysis of clinical practice underscore that selection of therapy in relapsed or refractory hematologic malignancies includes a need to weigh early, severe but more manageable toxicities against prolonged disease control and overall morbidity. CAR-T treatment has offered the prospect of long-term remission in patient populations who would otherwise be considered unlikely to respond to therapy, and its toxicity profile, albeit distinct, may be deemed tolerable, if not ideal, compared with the long-term toxicity profiles of other high-intensity therapies. This view is consistent with current therapies that emphasize individualized risk profiling rather than blanket judgments about treatment safety.
Limitations and Possible Biases. There are several limitations worth mentioning. First, the studies included differing in their definitions of toxicity, grading systems, and follow-up durations, particularly about late adverse events, leading to residual heterogeneity. Second, most analyses used aggregate data without patient-level adjustment for baseline risk predictors, such as comorbidity burden or immune reserve. Third, publication bias is not entirely avoidable since studies reporting severe or novel toxicities may have been preferentially included.
Furthermore, advances in CAR-T constructs and supportive care practices may limit the applicability of existing evidence in current clinical practice. Finally, comparing direct therapeutic findings with emerging drugs, including bispecific antibodies, has limited mature data, and reporting is heterogeneous, highlighting the need for further comparative effectiveness studies on bispecific antibody use.
Conclusions
In conclusion, our results show that CAR-T therapy carries a unique, comparator-dependent toxicity entity, not an overall toxicity load. Seen within the proper clinical and temporal context, CAR-T therapy can remain a significant and legitimate therapeutic option for selected patients with relapsed or refractory hematologic malignancies, provided delivery occurs in established centers equipped to mitigate its unique toxicity profile adequately.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Authors' contributions
Author ContributionsAll authors contributed equally to this work. Alexandra Porras-Ramírez (APR), Alejandro Rico-Mendoza (ARM), Liliana Encinales (LE), Oscar Madiedo (OM), Marianna Carrillo-Encinales (MCE), Juan Pablo Duque-Bolívar (JPDB), Juan Fernando Ramón-Cuellar (JFRC), Jorge Andrés Urquijo-Méndez (JAUM), Lucía Mora-Quiñones (LMQ), Alexandra Marín-Morales (AMM), Zuly Moreno-Perilla (ZMP), and Alejandra Hernández-Osorio (AHO) jointly participated in: study conception and design; development of the search strategy; literature screening and study selection; data extraction and verification; risk of bias assessment and grading of evidence; statistical analysis and interpretation; and drafting, reviewing, and editing of the manuscript.All authors read and approved the final manuscript and agree to be accountable for all aspects of the work.
Funding
This research received no funding or financial support from public, commercial, or not-for-profit organizations.
Data availability
All data analyzed in this meta-analysis were obtained from previously published and publicly available studies indexed in PubMed, Embase, Scopus, Web of Science, and the Cochrane Library.The detailed dataset used for analysis is available from the corresponding author upon reasonable request.
Declarations
Ethics approval and consent to participate
This systematic review and meta-analysis were based entirely on previously published data and did not involve any innovative studies with human participants or animals performed by the authors. Therefore, approval by an ethics committee or institutional review board (IRB) was not needed. All included studies were conducted by the ethical standards of their respective institutions and the Declaration of Helsinki and Good Clinical Practice guidelines. No identifiable individual data were used in this research.
Consent for publication
Not applicable. This study is a systematic review and meta-analysis based exclusively on previously published and aggregated data. No individual patient data, images, or other identifiable personal or clinical information were included in this manuscript.
Competing interests
The authors declare no competing interests.
References
1. Maude SL Laetsch TW Buechner J Tisagenlecleucel in children and young adults with B-cell lymphoblastic leukemia N Engl J Med 2018 378 5 439 48 10.1056/NEJMoa1709866 29385370 PMC5996391 Maude SL, Laetsch TW, Buechner J, et al. Tisagenlecleucel in children and young adults with B-cell lymphoblastic leukemia. N Engl J Med. 2018;378(5):439–48. 29385370 10.1056/NEJMoa1709866 PMC5996391
2. Schuster SJ Bishop MR Tam CS Tisagenlecleucel in adult relapsed or refractory diffuse large B-cell lymphoma N Engl J Med 2019 380 1 45 56 10.1056/NEJMoa1804980 30501490 Schuster SJ, Bishop MR, Tam CS, et al. Tisagenlecleucel in adult relapsed or refractory diffuse large B-cell lymphoma. N Engl J Med. 2019;380(1):45–56. 30501490 10.1056/NEJMoa1804980
3. Locke FL Neelapu SS Bartlett NL Long-term safety and activity of axicabtagene ciloleucel in refractory large B-cell lymphoma (ZUMA-1) Lancet Oncol 2019 20 1 31 42 10.1016/S1470-2045(18)30864-7 30518502 PMC6733402 Locke FL, Neelapu SS, Bartlett NL, et al. Long-term safety and activity of axicabtagene ciloleucel in refractory large B-cell lymphoma (ZUMA-1). Lancet Oncol. 2019;20(1):31–42. 30518502 10.1016/S1470-2045(18)30864-7 PMC6733402
4. Munshi NC Anderson LD Jr Shah N Idecabtagene vicleucel in relapsed and refractory multiple myeloma N Engl J Med 2021 384 8 705 16 10.1056/NEJMoa2024850 33626253 Munshi NC, Anderson LD Jr, Shah N, et al. Idecabtagene vicleucel in relapsed and refractory multiple myeloma. N Engl J Med. 2021;384(8):705–16. 33626253 10.1056/NEJMoa2024850
5. Brudno JN Kochenderfer JN Toxicities of chimeric antigen receptor T-cell therapy: recognition and management Blood 2016 127 26 3321 30 10.1182/blood-2016-04-703751 27207799 PMC4929924 Brudno JN, Kochenderfer JN. Toxicities of chimeric antigen receptor T-cell therapy: recognition and management. Blood. 2016;127(26):3321–30. 27207799 10.1182/blood-2016-04-703751 PMC4929924
6. Neelapu SS Tummala S Kebriaei P Chimeric antigen receptor T-cell therapy—assessment and management of toxicities Nat Rev Clin Oncol 2018 15 1 47 62 10.1038/nrclinonc.2017.148 28925994 PMC6733403 Neelapu SS, Tummala S, Kebriaei P, et al. Chimeric antigen receptor T-cell therapy—assessment and management of toxicities. Nat Rev Clin Oncol. 2018;15(1):47–62. 28925994 10.1038/nrclinonc.2017.148 PMC6733403
7. Lee DW Santomasso BD Locke FL ASTCT consensus grading for cytokine release syndrome and neurologic toxicity associated with immune effector cells Biol Blood Marrow Transpl 2019 25 4 625 38 10.1016/j.bbmt.2018.12.758 PMC12180426 30592986 Lee DW, Santomasso BD, Locke FL, et al. ASTCT consensus grading for cytokine release syndrome and neurologic toxicity associated with immune effector cells. Biol Blood Marrow Transpl. 2019;25(4):625–38. 10.1016/j.bbmt.2018.12.758 PMC12180426 30592986
8. Hill JA Li D Hay KA Infectious complications of CD19-targeted chimeric antigen receptor-modified T-cell immunotherapy Blood 2018 131 1 121 30 10.1182/blood-2017-07-793760 29038338 PMC5755046 Hill JA, Li D, Hay KA, et al. Infectious complications of CD19-targeted chimeric antigen receptor-modified T-cell immunotherapy. Blood. 2018;131(1):121–30. 29038338 10.1182/blood-2017-07-793760 PMC5755046
9. Cordeiro A Bezerra ED Hirayama AV Late events after treatment with CD19-targeted chimeric antigen receptor-modified T cells Biol Blood Marrow Transpl 2020 26 1 26 33 10.1016/j.bbmt.2019.08.003 PMC6953906 31419568 Cordeiro A, Bezerra ED, Hirayama AV, et al. Late events after treatment with CD19-targeted chimeric antigen receptor-modified T cells. Biol Blood Marrow Transpl. 2020;26(1):26–33. 10.1016/j.bbmt.2019.08.003 PMC6953906 31419568
10. Rejeski K Perez A Iacoboni G CAR-HEMATOTOX: a model for CAR T-cell–related hematotoxicity in large B-cell lymphoma Blood 2022 139 17 2438 51 Rejeski K, Perez A, Iacoboni G, et al. CAR-HEMATOTOX: a model for CAR T-cell–related hematotoxicity in large B-cell lymphoma. Blood. 2022;139(17):2438–51.
11. Santomasso BD Park JH Salloum D Clinical and biological correlates of neurotoxicity associated with CAR T-cell therapy Cancer Discov 2018 8 8 958 71 10.1158/2159-8290.CD-17-1319 29880584 PMC6385599 Santomasso BD, Park JH, Salloum D, et al. Clinical and biological correlates of neurotoxicity associated with CAR T-cell therapy. Cancer Discov. 2018;8(8):958–71. 29880584 10.1158/2159-8290.CD-17-1319 PMC6385599
12. Abramson JS Palomba ML Gordon LI Patient-reported outcomes in lymphoma patients receiving CAR T-cell therapy Blood Adv 2022 6 15 4475 84 Abramson JS, Palomba ML, Gordon LI, et al. Patient-reported outcomes in lymphoma patients receiving CAR T-cell therapy. Blood Adv. 2022;6(15):4475–84.
13. Palumbo A Anderson K Multiple myeloma N Engl J Med 2011 364 11 1046 60 10.1056/NEJMra1011442 21410373 Palumbo A, Anderson K. Multiple myeloma. N Engl J Med. 2011;364(11):1046–60. 21410373 10.1056/NEJMra1011442
14. June CH O’Connor RS Kawalekar OU CAR T cell immunotherapy for human cancer Science 2018 359 6382 1361 5 10.1126/science.aar6711 29567707 June CH, O’Connor RS, Kawalekar OU, et al. CAR T cell immunotherapy for human cancer. Science. 2018;359(6382):1361–5. 29567707 10.1126/science.aar6711
15. Fried S Avigdor A Bielorai B Early and late hematologic toxicity following CD19 CAR-T cells Bone Marrow Transpl 2019 54 10 1643 50 10.1038/s41409-019-0487-3 30809033 Fried S, Avigdor A, Bielorai B, et al. Early and late hematologic toxicity following CD19 CAR-T cells. Bone Marrow Transpl. 2019;54(10):1643–50. 10.1038/s41409-019-0487-3 30809033
16. Crump M Neelapu SS Farooq U Outcomes in refractory diffuse large B-cell lymphoma: results from SCHOLAR-1 Blood 2017 130 16 1800 8 10.1182/blood-2017-03-769620 28774879 PMC5649550 Crump M, Neelapu SS, Farooq U, et al. Outcomes in refractory diffuse large B-cell lymphoma: results from SCHOLAR-1. Blood. 2017;130(16):1800–8. 28774879 10.1182/blood-2017-03-769620 PMC5649550
17. Hutchings M Carlo-Stella C Bachy E Glofitamab step-up dosing induces high response rates in relapsed or refractory non-Hodgkin lymphoma J Clin Oncol 2021 39 18 1959 70 10.1200/JCO.20.03175 33739857 PMC8210975 Hutchings M, Carlo-Stella C, Bachy E, et al. Glofitamab step-up dosing induces high response rates in relapsed or refractory non-Hodgkin lymphoma. J Clin Oncol. 2021;39(18):1959–70. 33739857 10.1200/JCO.20.03175 PMC8210975
18. Viardot A Goebeler ME Hess G Phase 2 study of the bispecific antibody blinatumomab in relapsed/refractory B-cell malignancies Blood 2016 127 11 1410 6 10.1182/blood-2015-06-651380 26755709 PMC4797019 Viardot A, Goebeler ME, Hess G, et al. Phase 2 study of the bispecific antibody blinatumomab in relapsed/refractory B-cell malignancies. Blood. 2016;127(11):1410–6. 26755709 10.1182/blood-2015-06-651380 PMC4797019
19. Passweg JR Baldomero H Chabannon C Hematopoietic stem cell transplantation in Europe 2019: EBMT activity survey Bone Marrow Transpl 2021 56 7 1651 72 10.1038/s41409-021-01227-8 PMC8263343 33623153 Passweg JR, Baldomero H, Chabannon C, et al. Hematopoietic stem cell transplantation in Europe 2019: EBMT activity survey. Bone Marrow Transpl. 2021;56(7):1651–72. 10.1038/s41409-021-01227-8 PMC8263343 33623153
20. Shah NN Lee DW Yates B Long-term follow-up of CD19-CAR T-cell therapy Blood Adv 2022 6 7 2033 45 Shah NN, Lee DW, Yates B, et al. Long-term follow-up of CD19-CAR T-cell therapy. Blood Adv. 2022;6(7):2033–45.
21. Yakoub-Agha I Chabannon C Bader P Management of adults and children undergoing CAR-T-cell therapy: best practice recommendations Bone Marrow Transpl 2020 55 2 297 316 10.3324/haematol.2019.229781 PMC7012497 31753925 Yakoub-Agha I, Chabannon C, Bader P, et al. Management of adults and children undergoing CAR-T-cell therapy: best practice recommendations. Bone Marrow Transpl. 2020;55(2):297–316. 10.3324/haematol.2019.229781 PMC7012497 31753925