Treatment-Related Myelodysplastic Syndrome and Acute Myeloid Leukemia Associated with CAR T-Cell Therapies: A Real-World FAERS Pharmacovigilance Study

Abstract

Introduction: Chimeric antigen receptor (CAR) T-cell therapies have revolutionized treatment for relapsed/refractory hematologic malignancies, targeting CD19 in B-cell neoplasms and BCMA in multiple myeloma, with response rates exceeding 80%. However, long-term risks, including therapy-related myeloid neoplasms, such as myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML), are emerging 6–24 months post infusion, potentially linked to lymphodepleting chemotherapy, clonal hematopoiesis expansion, and inflammatory milieus. This FAERS pharmacovigilance analysis quantified MDS/AML reporting across seven FDA-approved CAR-T products to detect antigen-specific signals unattainable in pivotal trials with limited follow-up. Methods: Adverse event reports from FAERS (1 January 2013–10 February 2025) were queried for tisagenlecleucel, axicabtagene ciloleucel, brexucabtagene autoleucel, lisocabtagene maraleucel, obecabtagene autoleucel, idecabtagene vicleucel, and ciltacabtagene autoleucel, focusing on MedDRA terms for MDS/AML. Duplicates and ambiguous cases were excluded. Disproportionality was assessed using reporting odds ratios (RORs; lower 95% CI >1 signaling significance), comparing CAR-T-event pairs to database background, with subgroup analyses by antigen target. Results: Among 14,093,557 reports, CAR-T products linked to 303 MDS (brexucabtagene autoleucel ROR 97.93 [72.18–132.87], n = 44; axicabtagene ciloleucel ROR 58.70 [50.34–68.44], n = 172) and 129 AML cases (axicabtagene ciloleucel ROR 22.89 [18.23–28.73], n = 76). Signals were consistent across CD19- and BCMA-directed agents, absent only for recently approved obecabtagene autoleucel. Conclusions: CAR-T therapies exhibit disproportionate MDS/AML reporting in FAERS, supporting class-wide late hematologic toxicity in pretreated patients with clonal hematopoiesis. Enhanced surveillance, baseline profiling, and marrow evaluation for cytopenias are warranted, balancing curative benefits.

1. Introduction

Chimeric antigen receptor (CAR) T-cell therapies have fundamentally transformed the therapeutic paradigm for hematologic malignancies, achieving unprecedented rates of durable complete remission in patients with relapsed/refractory disease [1,2]. These autologous T lymphocytes, genetically reprogrammed to express synthetic receptors targeting tumor-associated antigens, predominantly CD19 in B-cell non-Hodgkin lymphoma (NHL), acute lymphoblastic leukemia (ALL), and chronic lymphocytic leukemia (CLL), and B-cell maturation antigen (BCMA) in multiple myeloma, mediate potent antigen-specific cytotoxicity through perforin/granzyme release and Fas/FasL signaling [3,4]. Pivotal trials have demonstrated objective response rates exceeding 80% and sustained remissions beyond three years in select cohorts, establishing CAR T-cell therapy as a curative modality for chemotherapy-refractory hematologic cancers [5,6,7,8].

However, the long-term safety profile remains incompletely defined as the first approved products reach their eighth year of clinical use. While acute toxicities such as cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS) are well-characterized and effectively mitigated, delayed hematologic sequelae have emerged as patients survive for years post infusion [9,10,11]. Among these, therapy-related myeloid neoplasms, particularly myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML), represent an concerning signal. Case series and registry analyses have documented MDS onset 6–24 months post CAR-T infusion, often with adverse cytogenetic profiles including complex karyotypes and TP53 aberrations incompatible with allogeneic hematopoietic stem cell transplantation [12,13,14].

Pathophysiologic mechanisms underlying CAR-T-associated myeloid neoplasms remain speculative but multifactorial. Lymphodepletion regimens incorporating fludarabine/cyclophosphamide expose hematopoietic stem cells to genotoxic insult, potentially compounded by radiation in prior consolidative therapy. Post-infusion marrow aplasia and prolonged cytopenia may enable clonal hematopoiesis of indeterminate potential (CHIP) expansion, with inflammatory cytokine milieus selecting for pre-leukemic clones harboring DNMT3A, TET2, or ASXL1 mutations. Critically, pivotal registration trials with median follow-up <36 months lacked power to detect rare late events occurring at incidences <1%, underscoring the need for post-marketing surveillance.

The FDA Adverse Event Reporting System (FAERS) provides a unique pharmacovigilance resource for detecting disproportionate reporting patterns indicative of emerging safety signals. With over 14,000,000 reports, FAERS enables detection of rare drug–event associations unattainable through prospective studies. This analysis systematically quantified MDS and AML reporting across seven FDA-approved CAR-T products, comparing CD19- versus BCMA-directed agents to elucidate antigen-specific risk patterns and inform long-term monitoring strategies.

2. Results

A disproportionality analysis using the FAERS database identified reports of MDS and AML associated with approved CAR T-cell therapies. Between database inception and the most recent FAERS update (Q3 2025), a total of 14,093,557 adverse event reports were registered, of which 8889 referenced MDS and 9758 referenced AML as preferred terms within MedDRA. Among these, CAR T-cell products collectively accounted for 303 MDS reports and 129 AML reports. Disproportionality signals, expressed as reporting odds ratios (RORs) with corresponding 95% confidence intervals (CIs), are summarized in

Table 1. Disproportionality analysis of myelodysplastic syndrome reports associated with CAR T-cell products in the FDA Adverse Event Reporting System database.

Drug	Target Antigen	Reporting Odds Ratio	Number of Reports
Tisagenlecleucel	CD19	28.43 (95% CI: 20.86, 38.75)	41
Axicabtagene Ciloleucel	CD19	58.70 (95% CI: 50.34, 68.44)	172
Brexucabtagene Autoleucel	CD19	97.93 (95% CI: 72.18, 132.87)	44
Lisocabtagene Maraleucel	CD19	19.65 (95% CI: 7.33, 52.68)	4
Obecabtagene Autoleucel	CD19	N/a	0
Idecabtagene Vicleucel	BCMA	35.37 (95% CI: 17.55, 71.29)	8
Ciltacabtagene Autoleucel	BCMA	17.67 (95% CI: 12.60, 24.79)	34

and

Table 2. Disproportionality analysis of acute myeloid leukemia reports associated with CAR T-cell products in the FDA Adverse Event Reporting System database.

Drug	Target Antigen	Reporting Odds Ratio	Number of Reports
Tisagenlecleucel	CD19	11.86 (95% CI: 7.55, 18.63)	19
Axicabtagene Ciloleucel	CD19	22.89 (95% CI: 18.23, 28.73)	76
Brexucabtagene Autoleucel	CD19	17.33 (95% CI: 8.98, 33.44)	9
Lisocabtagene Maraleucel	CD19	8.89 (95% CI: 2.21, 35.71)	2
Obecabtagene Autoleucel	CD19	N/a	0
Idecabtagene Vicleucel	BCMA	11.91 (95% CI: 3.82, 37.10)	3
Ciltacabtagene Autoleucel	BCMA	9.41 (95% CI: 6.06, 14.61)	20

.

2.1. Myelodysplastic Syndrome

MDS reporting was notably enriched across CD19- and BCMA-directed CAR T-cell constructs. Among anti-CD19 agents, axicabtagene ciloleucel demonstrated the highest signal strength (ROR = 58.70; 95% CI 50.34–68.44; n = 172), followed by brexucabtagene autoleucel (ROR = 97.93; 95% CI 72.18–132.87; n = 44) and tisagenlecleucel (ROR = 28.43; 95% CI 20.86–38.75; n = 41). In contrast, lisocabtagene maraleucel yielded a smaller, though still elevated, signal (ROR = 19.65; 95% CI 7.33–52.68; n = 4), while no MDS cases were identified for obecabtagene autoleucel, likely reflecting its recent regulatory approval and limited post-marketing duration.

For BCMA-targeted therapies, MDS was reported in association with idecabtagene vicleucel (ROR = 35.37; 95% CI 17.55–71.29; n = 8) and ciltacabtagene autoleucel (ROR = 17.67; 95% CI 12.60–24.79; n = 34). Overall, these findings suggest that MDS, while infrequent in absolute numbers, occurs at a significantly elevated frequency relative to background reporting within FAERS.

2.2. Acute Myeloid Leukemia

Similarly, AML reports were detected across multiple CAR T-cell constructs, involving both CD19- and BCMA-targeted products. Among CD19-directed therapies, axicabtagene ciloleucel (ROR = 22.89; 95% CI 18.23–28.73; n = 76) and brexucabtagene autoleucel (ROR = 17.33; 95% CI 8.98–33.44; n = 9) demonstrated the strongest disproportionality signals, while tisagenlecleucel exhibited a modest but significant elevation (ROR = 11.86; 95% CI 7.55–18.63; n = 19). Lisocabtagene maraleucel (ROR = 8.89; 95% CI 2.21–35.71; n = 2) produced limited case counts, and obecabtagene autoleucel again showed no AML reports. Among BCMA-related products, idecabtagene vicleucel (ROR = 11.91; 95% CI 3.82–37.10; n = 3) and ciltacabtagene autoleucel (ROR = 9.41; 95% CI 6.06–14.61; n = 20) also displayed elevated AML signal estimates. Collectively, these observations indicate that secondary AML, like MDS, represents a reproducible signal emerging across multiple CAR T-cell platforms within the FAERS dataset.

3. Discussion

In this large FAERS-based pharmacovigilance analysis, CAR T-cell products showed consistent disproportionality signals for both MDS and AML across CD19- and BCMA-directed constructs, reinforcing an emerging body of case-based and cohort evidence that therapy-related myeloid neoplasms constitute a real, albeit infrequent, late toxicity of CAR T-cell therapy. The elevated RORs observed for multiple products, together with the absence of signal only for the most recently approved construct, align with the temporal pattern seen in clinical reports, in which myeloid neoplasms typically manifest months to years after infusion, once sufficient exposure time has accrued. When interpreted alongside the published literature, these findings support the concept that CAR T-cell therapy does not act in isolation, but rather operates within a pre-injured, genomically primed hematopoietic milieu shaped by prior cytotoxic therapy and underlying clonal hematopoiesis.

3.1. Reports of Myelodysplastic Syndrome After CAR T-Cell Therapy

Several detailed case reports have described high-risk MDS arising after CD19-directed CAR T-cell therapy for B-cell lymphomas, generally in the setting of prolonged post-infusion cytopenias. A 2023 report described a patient with relapsed DLBCL who developed high-risk MDS eight months after CAR-T therapy, with multilineage dysplasia, monosomy 7, and a RUNX1 mutation emerging on a background of pre-existing clonal hematopoiesis involving CSF3R and CEBPA [12]. This case mirrors the latency and adverse cytogenetic profiles expected for classic therapy-related MDS, suggesting that CAR T-cell therapy is superimposed on an already compromised stem cell compartment rather than acting as a sole inciting event.

A case series summarized four patients with relapsed/refractory B-cell lymphomas treated with CD19-directed CAR-T who subsequently developed MDS or clonal cytopenias of undetermined significance (CCUS) [13]. The median age was 72.5 years and the median number of prior lines of therapy was three, highlighting a heavily pretreated, older population. Two patients manifested CCUS within 1–2 months of infusion, whereas two developed frank MDS at 10 and 26 months, illustrating a spectrum from early clonal cytopenias to overt myeloid neoplasia. Notably, pre-CAR-T assessments did not reveal dysplastic clones, underscoring the difficulty of distinguishing truly de novo events from expansion of below-detection clones in routine practice.

Additional reports in B-ALL further support the occurrence of myeloid malignancies after CAR-T, though often in complex clinical contexts. A JITC series described two multiply relapsed B-ALL patients who, following CAR-T, developed myeloid malignancies: one with myeloid sarcoma in the setting of prior MDS, and another with donor-derived MDS with monosomy 7 after allogeneic transplantation, later evolving to ambiguous lineage T/myeloid acute leukemia [15]. These vignettes emphasize that post-CAR-T myeloid neoplasms frequently arise in patients with pre-existing myeloid disease, prior transplantation, or lineage plasticity, complicating causal attribution yet reinforcing the need for systematic marrow evaluation in atypical relapses.

3.2. BCMA-Directed CAR-T and Clonal Myeloid Dysplasia

The signal observed in FAERS for BCMA-directed products is concordant with emerging data in multiple myeloma cohorts. A 2023 study evaluated five myeloma patients with MDS or clonal myeloid dysplasia in the context of anti-BCMA CAR-T [16]. Four developed MDS only after CAR-T, whereas one had established MDS before infusion; importantly, deep sequencing of serial samples demonstrated that all MDS-related mutations were already present prior to CAR-T in every patient, with no novel driver variants emerging post-infusion. These data suggest that BCMA-directed CAR-T primarily promotes expansion or unmasking of pre-existing MDS clones rather than inducing de novo genomic lesions, a mechanistic model that dovetails with the high RORs in FAERS yet cautions against attributing causality solely to the cellular product.

The myeloma population is particularly enriched for therapy-related clonal evolution given exposure to alkylators, proteasome inhibitors, immunomodulatory agents, and autologous transplantation. The series noted that the BCMA CAR-T-treated patients with MDS had extensive prior therapy and that several harbored high-risk mutations such as TP53 and spliceosome lesions. In such patients, the profound lymphodepletion, inflammatory milieu, and transient marrow aplasia characteristic of CAR-T may provide a permissive niche for pre-existing myeloid clones to gain a competitive advantage, consistent with broader models of therapy-selected clonal hematopoiesis [16,17,18].

3.3. Reports of Acute Myeloid Leukemia After CAR T-Cell Therapy

Compared with MDS, AML after CAR-T remains less frequently described but is increasingly reported. A 2024 case report detailed a patient with relapsed mantle cell lymphoma who developed acute erythroid leukemia with biallelic TP53 inactivation 26 months after brexucabtagene autoleucel; the patient had no evidence of bone marrow pathology prior to lymphoma-directed treatments, underscoring a late-emerging, therapy-related phenotype [14]. The JITC B-ALL series also illustrates that evolution to myeloid leukemia post-CAR-T may occur via lineage switch, particularly in the presence of donor-derived or antecedent myeloid clones [15]. Such cases blur the boundary between “secondary AML” and transformed antecedent disease and highlight that not all post-CAR-T myeloid neoplasms are straightforward t-MDS/t-AML as traditionally defined. Nonetheless, these observations are consistent with the FAERS AML signals across multiple constructs and suggest that CAR-T-treated patients represent a population at meaningful risk for subsequent myeloid transformation.

3.4. Real-World Cohort Data

Case reports and small series are inherently subject to publication bias; thus, real-world cohort data are critical for placing the FAERS disproportionality signals into quantitative context. The ClonHema study recently reported on 556 patients receiving CD19 CAR-T across 16 Italian centers, identifying 14 secondary myeloid neoplasms (13 MDS and 1 AML), with a median time to SMN of 7.9 months and cumulative incidence of 1.1%, 1.8%, and 2.7% at 6 months, 1 year, and 4 years, respectively [19]. Although the absolute incidence was low, it is notable in a population with otherwise limited long-term survival options and compares unfavorably to background myeloid neoplasm rates in similar age groups.

Complementary data from a recent letter identified 20 myeloid neoplasms among adults treated with CAR-T for lymphoproliferative disease or myeloma, with a median time to event of 10 months and cumulative incidence of 4%, 6%, and 9% at 1, 2, and 3 years [20]. In that cohort, clonal hematopoiesis was present in 64% of evaluable cases, frequently involving TP53 and PPM1D, and baseline factors such as age ≥65 and thrombocytopenia were strongly associated with subsequent myeloid neoplasms. Together, these cohort data suggest that the FAERS ROR elevations are not spurious but reflect a clinically meaningful risk concentrated in older, heavily pretreated patients with pre-existing clonal or cytopenic abnormalities.

3.5. Mechanistic Considerations

The convergence of case reports, myeloma series, and cohort data strongly implicates clonal hematopoiesis as a central mediator linking CAR-T exposure and therapy-related myeloid neoplasms [17,18,19,20]. Reviews of therapy-selected clonal hematopoiesis demonstrate that cytotoxic and radiation therapies can preferentially expand clones with mutations in DNA damage response (e.g., TP53, PPM1D), chromatin modifiers (e.g., ASXL1), and epigenetic regulators (e.g., DNMT3A, TET2), thereby seeding a reservoir of pre-malignant cells primed for subsequent transformation. In the CAR-T setting, several groups have documented high rates of clonal hematopoiesis at baseline and in patients who later develop myeloid neoplasms.

Moreover, a recent analysis presented at ASH linked clonal hematopoiesis to higher rates of severe CRS and prolonged cytopenias following CD19 and BCMA CAR-T, suggesting that inflammatory cytokine milieus and impaired hematopoietic reserve may act synergistically to promote clonal expansion and marrow failure [21]. These observations fit well with the BCMA CAR-T series showing that MDS clones and their full mutational complement were detectable pre-CAR-T in all patients, with CAR-T functioning as a “permissive accelerator” rather than a primary mutagen [16]. Mechanistically, lymphodepletion with fludarabine/cyclophosphamide, subsequent profound cytopenias, and repeated inflammatory insults may create a selective environment that favors survival and outgrowth of pre-existing, therapy-hardened myeloid clones.

Within this framework, the elevated RORs observed across CAR-T products in FAERS likely reflect an intersection of three components: (1) a high burden of prior cytotoxic exposure and associated clonal hematopoiesis, (2) an acute phase of marrow suppression and cytokine-driven stress following CAR-T, and (3) improved long-term survival that allows latent myeloid neoplasms to clinically manifest [14,19]. This model is consistent with the adverse cytogenetic features (e.g., monosomy 7, TP53) and relatively short latency (often <12–24 months) reported across cases and series.

3.6. Interpretation of FAERS Disproportionality Signals

Within this broader context, the FAERS findings can be interpreted as confirmatory of a signal already hinted at by the clinical literature rather than as a de novo discovery. The elevated RORs for both MDS and AML across CD19-directed products, particularly axicabtagene ciloleucel and brexucabtagene autoleucel, are congruent with their extensive real-world uptake in heavily pretreated lymphoma populations and the case series focusing on these constructs. Similarly, the presence of significant MDS and AML signals for BCMA-directed agents mirrors observations from multiple myeloma cohorts in which pre-existing clonal myeloid lesions are common and may expand following CAR-T therapy. The absence of signal for more recently approved products in FAERS is most parsimoniously explained by limited follow-up and fewer exposed patients rather than superior intrinsic safety.

However, several caveats are essential. FAERS is a spontaneous reporting system subject to underreporting, stimulated reporting after safety communications, and incomplete clinical annotation. RORs reflect disproportionality rather than incidence and cannot distinguish causal from confounded associations; they are particularly vulnerable to confounding by indication and prior therapy, both of which are profound in the CAR-T-eligible population. Because FAERS lacks systematic data on baseline clonal hematopoiesis, prior transplantation, and cumulative chemotherapy dose, the extent to which the observed signals are attributable to CAR-T itself versus antecedent exposures cannot be quantified.

3.7. Clinical Implications for Surveillance and Diagnostic Workup

Despite these limitations, convergent evidence from FAERS, case reports, and cohort studies has direct implications for clinical practice. Multiple reports now emphasize that persistent or late-onset cytopenias after CAR-T, beyond the expected 30–90-day recovery window, should prompt bone marrow evaluation rather than being dismissed as “slow count recovery.” Cohort data from suggest that baseline factors such as age, platelet count, and clonal hematopoiesis can stratify risk of post-CAR-T myeloid neoplasms, potentially informing pre-infusion counseling and surveillance intensity. Some authors have advocated for baseline NGS profiling of peripheral blood or marrow in CAR-T candidates, particularly older patients and those with extensive prior therapy, to identify high-risk CH or CCUS and to guide discussions about the small but non-trivial risk of t-MDS/t-AML. While routine CH screening is not yet standard of care, the accumulating evidence, together with pharmacovigilance signals, suggests that this may become an important component of comprehensive CAR-T survivorship care.

3.8. Strengths and Limitations

The principal strength of the present FAERS-based study is its ability to capture rare, delayed events across the entire marketed CAR-T portfolio, including both CD19 and BCMA constructs, in a way that complements the relatively small, single-center series currently dominating the literature. By using standardized MedDRA terms and established disproportionality methodology, the analysis provides a product-level overview that would be difficult to obtain from trial or institutional data alone. Moreover, the observation of consistent signals for both MDS and AML across multiple independent constructs strengthens the argument that therapy-related myeloid neoplasms are a class effect within a specific high-risk population, rather than isolated idiosyncratic events. Conversely, the limitations intrinsic to FAERS are substantial and must temper causal inferences. There is no reliable denominator of exposed patients for each CAR-T product, precluding true incidence estimation and comparison of absolute risks between constructs. Clinical details such as prior lines of therapy, transplant history, radiation exposure, baseline CH status, and cytogenetics are incompletely reported, limiting mechanistic insights and confounder adjustment. Misclassification of events (e.g., distinguishing de novo AML from transformed MDS or lineage-switched ALL) is likely, especially when detailed pathology reports are not available. Finally, reporting may be biased toward more severe or unusual cases, and heightened regulatory attention to secondary malignancies after CAR-T may itself stimulate increased reporting.

4. Methods

This study systematically extracted adverse event data from the U.S. Food and Drug Administration Adverse Event Reporting System (FAERS), covering the period from 1 January 2013 to 10 February 2025. The methods employed were consistent with previously validated approaches for FAERS-based signal detection [22,23,24]. The FAERS database constitutes a comprehensive, publicly accessible pharmacovigilance repository compiling spontaneously reported adverse events associated with pharmaceutical and biologic products from healthcare professionals, patients, and manufacturers worldwide.

4.1. Data Extraction

For this analysis, reports linked to all FDA-approved CAR T-cell therapies were included based on both their generic names: tisagenlecleucel, axicabtagene ciloleucel, brexucabtagene autoleucel, lisocabtagene maraleucel, obecabtagene autoleucel, idecabtagene vicleucel, and ciltacabtagene autoleucel. Data extraction was conducted using OpenVigil 2.1 [25,26], a validated pharmacovigilance query interface that enables structured interrogation of FAERS data and automatic duplicate exclusion. Adverse events were identified and categorized using Medical Dictionary for Regulatory Activities (MedDRA) preferred terms, focusing on hematologic malignancies, specifically myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML), within the hematopoietic and lymphoid tissue system organ class. Duplicate entries, reports missing drug identifiers, or events with ambiguous causality were systematically excluded via OpenVigil 2.1.

4.2. Disproportionality Analysis

Potential safety signals were assessed through disproportionality analysis, a standard statistical approach in pharmacovigilance that compares the observed frequency of a drug–event combination to its expected reporting rate within the overall FAERS database. This method identifies instances where an adverse event occurs more frequently than predicted by chance, suggesting a possible drug-related association warranting further evaluation. Consistent with FDA pharmacovigilance convention, disproportionality was quantified using the Reporting Odds Ratio (ROR) with corresponding 95% confidence intervals (CIs), defined as: ROR= (a/c)/(b/d), where a represents the number of reports containing both the CAR-T product and the event of interest; b, the number of reports including the CAR-T product without the event; c, the number of reports containing the event without the CAR-T product; and d, all remaining reports. A statistically significant signal was defined a priori as an ROR with the lower bound of the 95% CI > 1.00, consistent with accepted pharmacovigilance thresholds. Comparative ROR analyses were then performed across CD19- and BCMA-directed CAR T-cell agents to evaluate for antigen-specific variance in hematologic risk patterns.

4.3. Methodological and Ethical Standards

This study adhered to the REporting of studies Conducted using Observational Routinely collected Data (RECORD) statement, which extends the STROBE guidelines to enhance transparency and reproducibility in real-world data research. Analyses were conducted using the complete FAERS dataset without patient-level identifiers or protected health information, ensuring compliance with established ethical and methodological standards for secondary pharmacovigilance research. FAERS data are publicly available and fully de-identified; therefore, institutional ethics board approval and informed consent were not required for this analysis.

5. Conclusions

Taken together, the FAERS disproportionality signals and the growing body of case reports, small series, and real-world cohorts support the conclusion that secondary MDS and AML are genuine, though uncommon, late complications of CAR T-cell therapy in heavily pretreated patients with lymphoid malignancies and myeloma. The available data are most consistent with a model in which CAR-T acts as a permissive accelerator of pre-existing clonal hematopoiesis and therapy-induced genomic injury, rather than as a de novo leukemogenic agent. In this context, vigilant long-term hematologic surveillance, early marrow assessment of persistent cytopenias, and consideration of baseline CH profiling appear warranted, while acknowledging that the absolute benefit of CAR-T in terms of survival and disease control remains substantial for appropriately selected patients.