T-ICAHT: grading and prognostic impact of thrombocytopenia after CAR T-cell therapy Open Access

Although chimeric antigen receptor (CAR) T-cell (CAR-T) therapy represents a significant advance for the treatment of advanced B-cell malignancies, it still faces significant side effects.¹ Both cytokine release syndrome (CRS) and immune effector cell–associated neurotoxicity syndrome (ICANS) were swiftly recognized as prototypical complications of CAR-T therapy, and efforts have since been made to homogenize definitions, grading, and management protocols.² More recently, hematological toxicity has been recognized as the most common high-grade CAR-T–related adverse event.³ Cytopenias are observed irrespective of the CAR-T product, target antigen, and disease entity. Patients typically present with various degrees of (pan)cytopenia, which can be profound and long lasting.^4,5 Importantly, cytopenias represent an important contributor to morbidity and mortality, particularly due to neutropenia-mediated infectious complications and, to a lesser extent, bleeding-related complications.^6,7 

In an effort to standardize the assessment of cytopenias after CAR-T therapy, the European Hematology Association and European Society for Blood and Marrow Transplantation (EBMT) defined immune effector cell–associated hematotoxicity (ICAHT) as a distinct toxicity category.^8,9 They introduced a consensus-based grading system (grades 1-4) resting solely on the depth and duration of neutropenia (hereafter termed N-ICAHT), separating early (days 0-30) from late (after day 30) manifestations.⁹ In subsequent studies, this N-ICAHT grading was shown to be a reliable tool to risk stratify patients for severe infections, nonrelapse mortality (NRM), and survival.^10-12 Owing to the nature of the grading system, these initial studies have predominantly focused on the prognostic significance of neutropenia and omitted other cytopenia variations such as thrombocytopenia and anemia. Recently, deliberations by an expert panel of the American Society for Transplantation and Cellular Therapy (ASTCT) on CAR-related toxicities (Washington DC, November 2024) urged an evidence-based approach to guide the decision regarding inclusion vs omission of thrombocytopenia, in particular for the forthcoming revised consensus grading (F. Locke, M. Maus, M.A. Perales, personal communication, 12 February 2025). The group highlighted the clinical importance of repeated hospitalizations due to transfusion dependence and the potential risk of significant bleeding complications due to protracted thrombocytopenia.

The primary objective of this study was to provide such an evidence base in a large multicenter cohort spanning multiple CAR-T products across various advanced B-cell malignancies and age groups. To this end, we specifically aimed to delineate the dynamics of thrombocytopenia after CAR-T therapy, assess the degree of overlap with neutropenia, and evaluate the impact of thrombocytopenia on key clinical outcomes such as transfusion utilization and survival. To enable seamless integration into the existing ICAHT framework, we developed a novel grading system termed T-ICAHT, with thresholds that align closely with N-ICAHT.

This multicenter observational retrospective study included patients treated between April 2016 and November 2024 at Memorial Sloan Kettering (MSK) Cancer Center (New York, NY), Rambam Health Care Campus (Haifa, Israel), Sheba Medical Center (Ramat Gan, Israel), and the Ludwig-Maximilian-University Hospital (Munich, Germany). The T-ICAHT grading system was first developed in a cohort of 744 adult patients (age ≥18 years) with a diagnosis of mature B-cell malignancies, including large B-cell lymphoma (LBCL), follicular lymphoma (FL), and mantle cell lymphoma (MCL), receiving one of the following CD19-directed commercial CAR-T products: axicabtagene ciloleucel (axi-cel), tisagenlecleucel (tisa-cel), lisocabtagene maraleucel (liso-cel), or brexucabtagene autoleucel (brexu-cel). All patients without longitudinal platelet counts or those treated with noncommercially available CAR-T products were excluded. Patient demographic data were captured in institutional REDCap databases.¹³ 

The grading system was further validated in 3 independent cohorts including: (1) patients with relapsed/refractory (R/R) multiple myeloma (MM) receiving idecabtagene vicleucel (ide-cel), ciltacabtagene autoleucel, and orvacabtagene autoleucel (JCARH125)¹⁴; (2) patients treated with an academic point-of-care CD28z-based product for R/R LBCL at Sheba Medical Center¹⁵; and (3) pediatric and adult patients with B-cell acute lymphoblastic leukemia (B-ALL) treated with commercial and investigational CD19 CAR-T products.¹⁶ The institutional review boards of participating institutions approved the study in accordance with the Declaration of Helsinki, and all patients signed informed consent for treatment.

Early T-ICAHT was modeled after N-ICAHT⁹ and graded on a 4-grade scale based on the cumulative depth and duration of thrombocytopenia within the first 30 days after infusion (Table 1). Late T-ICAHT was defined based on the depth of thrombocytopenia and timing from CAR-T infusion (days 31-100). N-ICAHT was graded according to European Hematology Association/EBMT consensus grading.⁹ A fully automated computational pipeline was developed to enable continuous and daily assessment of T-ICAHT and N-ICAHT grades (code developed by T.F. and available in the GitHub repository; https://github.com/Shouval-Lab/T-ICAHT-Grading).

CRS and ICANS were graded according to ASTCT consensus definitions.² The CAR-HEMATOTOX score was calculated and classified into risk groups as previously described (low, score 0-1; high, score ≥2).²⁰ For MSK Cancer Center patients (n = 462), all bleeding episodes until day 100 were screened using the International Classification of Diseases (ICD-10) codes and then reviewed to assess their severity according to the Common Terminology Criteria for Adverse Events version 5.0. Major bleeds were defined as grade ≥3 bleeding episodes. Overall survival (OS) was defined as the time from CAR-T infusion to death from all causes. Progression-free survival (PFS) was defined as the time from CAR-T infusion to death or relapse/progression, whichever came first. Relapse/progression was assessed at the individual institutions according to the Lugano criteria using positron emission tomography–computed tomography.²¹ NRM was defined as death from any cause without prior relapse/progression.

Categorical variables were described by frequency and percentage, and continuous variables were described by median and range. Fisher exact and Kruskal-Wallis tests were used to study categorical and continuous variables, respectively. OS and PFS were estimated using the Kaplan-Meier method,²² and log-rank tests were used for comparisons. Survival probabilities at 2 years are given as percentages and 95% confidence intervals (95% CIs). Cumulative incidence functions were used to estimate relapse/progression incidence and NRM, and comparisons were made using the Gray test.^23,24 Competing risks were death for relapse/progression incidence and relapse/progression for NRM. Landmark analysis was performed at day 30 and day 100 to evaluate the impact of early and late T-ICAHT on outcomes, respectively. Multivariate analyses were performed using the Cox proportional hazards model, including potential confounding factors.²⁵ All P values were 2-sided, and P value <.05 was considered statistically significant. Data were analyzed using R (version 4.1.2).

The baseline characteristics of the 744 patients of the core NHL cohort are summarized in Table 2. The median age was 65 years, with 64% of patients being male. LBCL was the most common diagnosis (85%). At the time of lymphodepletion, 40% had elevated lactate dehydrogenase (LDH), and 38% had a high CAR-HEMATOTOX score. Axi-cel was the predominant CAR-T product, administered to 48% of patients, followed by tisa-cel (23%), liso-cel (22%), and brexu-cel (7%). Bridging therapy was used in 78% of cases.

Higher early T-ICAHT grades were associated with lower Karnofsky performance status (KPS), MCL diagnosis, and brexu-cel treatment. Patients with higher early T-ICAHT grades also had lower median platelet counts before lymphodepletion and exhibited more adverse disease characteristics, including a greater number of prior therapy lines and elevated lactate dehydrogenase levels. Conversely, FL diagnosis and liso-cel administration were more frequent in patients with lower T-ICAHT grades.

Trends of platelet counts from the time of lymphodepletion until day 100 are shown in Figure 1A, with an early lymphodepletion-associated nadir observed around day 5, followed by an initial recovery/plateau phase. This was followed by a notable second nadir occurring around day 30 after infusion, followed by slow count normalization beginning in month 2. Overall, 43.5% and 22.6% of patients met a thrombocytopenia threshold of 50 × 10⁹/L and 20 × 10⁹/L, respectively. The dynamics of platelet count recovery varied according to early T-ICAHT grades (Figure 1B). A clear biphasic or “double-dip” pattern was noted in the patients with grade 0 and 1 early T-ICAHT. Conversely, patients with grades 2 and 3 early T-ICAHT displayed only moderate and short recovery, with the second nadir occurring at lower counts than the first. Finally, grade 4 early T-ICAHT was characterized by an “aplastic” pattern,^20,28 with absent count recovery after lymphodepletion.

Overall, 324 patients (43%) developed early T-ICAHT (days 0-30), with the following grade distribution: 10% grade 1; 11% grade 2; 11% grade 3; and 12% grade 4 (Figure 2A). Late T-ICAHT (days 31-100) was observed in 229 of 550 evaluable patients (42%), with 18% being grade 1, 11% grade 2, 5.6% grade 3, and 7.5% grade 4 (Figure 2A). Although most patients retained their T-ICAHT grade from early to late phases, some transitions were observed (Figure 2B), including 10% of patients who progressed from early grade 0 to late grade 1 to 2.

Early and late grades of T-ICAHT and N-ICAHT were generally correlated, although notable discordance was observed (Figure 2C-D; supplemental Table 1, available on the Blood website). Among patients without early T-ICAHT, 55% had early N-ICAHT grades 1 to 2, whereas 18% had grades 3 to 4. Conversely, among those with early T-ICAHT grades 3 to 4, up to 55% of patients had early N-ICAHT grades 3 to 4, whereas 40% had N-ICAHT grades 1 to 2 (Figure 2C). For late T-ICAHT, 30% of patients without late T-ICAHT had late N-ICAHT grades 1 to 2, and 3% had grades 3 to 4. Meanwhile, among those with late T-ICAHT grades 3 to 4, up to 67% had late N-ICAHT grades 3 to 4, whereas 26% had N-ICAHT grades 1 to 2 (Figure 2D).

On multivariable analysis (Figure 3A; supplemental Table 2), early severe (grade ≥3) T-ICAHT was associated with a lower Karnofsky performance status (adjusted odds ratio [aOR], 2.0; 95% CI, 1.2-3.4), a high HEMATOTOX score (aOR, 6.6; 95% CI, 4.2-10.4), and the administration of bridging therapy (aOR, 2.1; 95% CI, 1.2-4.0). In contrast, patients with FL had a lower risk of early severe T-ICAHT (aOR, 0.12; 95% CI, 0.01-0.6). For late severe (grade ≥3) T-ICAHT, we noted similar associations with performance status, HEMATOTOX, and bridging therapy (supplemental Table 2). Among patients with a high baseline HEMATOTOX score (≥2), we noted a significantly higher proportion of patients who went on to develop severe early (45% vs 9%; P < .001) and severe late T-ICAHT (28% vs 5%; P < .001) than patients with a low-risk score (Figure 3B). Of interest, early T-ICAHT severity also correlated with more pronounced immunotoxicity, as reflected by higher CRS and ICANS grades (supplemental Table 3).

In 462 evaluable patients (MSK cohort), we noted an increased bleeding rate, including both all-grade (P = .011) and major bleeding events (P = .024), in patients with higher T-ICAHT grades (supplemental Table 3). No fatal bleeds were observed. Platelet and packed red blood cell (RBC) utilization successively increased with T-ICAHT severity (Figure 4A-B). For example, 59% of patients with grade 4 early T-ICAHT received ≥5 platelet transfusions during the first 30 days (Figure 4A), highlighting their profound transfusion dependence. In general, we noted a strong temporal association between T-ICAHT grades and the use of platelet transfusions (supplemental Figure 1). Furthermore, patients with grade 4 early T-ICAHT also showed notable codependence for RBC transfusions, with 45% receiving >5 RBC transfusions during the first 30 days (Figure 4B). Concomitantly, the use of thrombopoietin (TPO) mimetics was more common in patients with higher-grade early T-ICAHT (grade 0, 0%; grade 1-2, 3%; grade ≥3, 28%; P < .001; supplemental Table 3). This was particularly evident in patients with grade 4 early T-ICAHT, of whom 47% received TPO mimetics.

After landmarking at day 30 after CAR-T, early T-ICAHT grades were inversely correlated with survival outcomes, with 2-year OS estimates ranging from 67% (grade 0) to 48% (grade 1-2) and 35% (grade ≥3; log-rank P < .001; Figure 4C) and 2-year PFS estimates of 51% (grade 0), 41% (grade 1-2), and 31% (grade ≥3; P < .001; Figure 4D). In a multivariable Cox regression analysis accounting for key patient-, disease-, and treatment-related features (Figure 4E; supplemental Table 4), the independent prognostic capacity of early T-ICAHT for OS was confirmed for grades 1 to 2 (adjusted hazard ratio [aHR], 1.4; 95% CI, 1.1-1.9) and grades 3 to 4 (aHR, 2.0; 95% CI, 1.5-2.7). On multivariable analysis for PFS (Figure 4F; supplemental Table 4), an association was only observed for grade 3 to 4 early T-ICAHT (aHR, 1.5; 95% CI, 1.2-2.0). When early N-ICAHT was included in multivariable Cox regression models, early T-ICAHT maintained its prognostic value for OS and PFS (supplemental Table 5). The survival disadvantage of grade ≥3 early T-ICAHT was predominantly driven by an increased relapse risk (supplemental Table 6). Concomitantly, the 2-year cumulative incidence of relapse was 44% in grade 0, 47% in grades 1 to 2, and 60% in grade ≥3 (P = .003), whereas the 2-year cumulative incidence of NRM was 6.3% in grade 0, 12% in grades 1 to 2, and 9.4% in grade ≥3 (P = .038).

After landmarking at day 100 after infusion, late T-ICAHT grades were also inversely associated with survival outcomes, with 2-year OS estimates ranging from 72% (grade 0) to 51% (grade 1-2) and 48% (grade ≥3; P < .001; supplemental Figure 2A); and 2-year PFS estimates of 63% (grade 0), 51% (grade 1-2), and 46% (grade ≥3; P = .003; supplemental Figure 2B). In multivariable Cox regression analysis (supplemental Figure 2C; supplemental Table 7), we confirmed the independent prognostic capacity of grade 1 to 2 (aHR, 1.8; 95% CI, 1.2-2.5) and grade 3 to 4 late T-ICAHT for OS (aHR, 1.7; 95% CI, 1.1-2.8). In addition, both grade 1 to 2 (aHR, 1.5; 95% CI, 1.01-2.1) and grade 3 to 4 late T-ICAHT (aHR, 1.9; 95% CI, 1.1-3.2) were independently associated with PFS (supplemental Figure 2D; supplemental Table 7). When late N-ICAHT was included in multivariable Cox regression models, late T-ICAHT also maintained its prognostic value for OS and PFS (supplemental Table 8). Again, the survival disadvantage of late T-ICAHT was predominantly driven by an increased relapse risk (supplemental Table 6).

Additionally, we evaluated the predictive capacity of a composite ICAHT grading system based on the maximum grade of either N-ICAHT or T-ICAHT. For composite early ICAHT, only grades 3 to 4 were associated with OS but not with PFS (supplemental Table 9). Similarly, composite late ICAHT grades 1 to 2 and 3 to 4 were associated with OS but not with PFS (supplemental Table 10). Thus, the composite grading system demonstrated reduced ability to differentiate survival outcomes compared to the separated ICAHT grading.

To evaluate the broad applicability of the novel T-ICAHT grading system, we assessed its clinical and prognostic utility in 3 independent cohorts (Figure 5; supplemental Tables 11-13).

For the Sheba NHL cohort, 135 patients received an academic point-of-care CD28z-based product. The median age was 53 years; 40% of patients were female; and most patients (69%) had received at least 3 lines of prior therapy (supplemental Table 11). The median prelymphodepletion platelet count was 164 × 10⁹/L. In contrast to the core NHL cohort, only 25 patients (19%) received bridging therapy. We confirmed the association between higher T-ICAHT grades and elevated baseline HEMATOTOX scores, poor performance status, and use of bridging therapy.

Regarding the grade distribution of early T-ICAHT, 67% were grade 0, 15% had grades 1 to 2, and 18% had grades 3 to 4 (Figure 5A). Early T-ICAHT grades were associated with OS, with 2-year OS estimates ranging from 64% (grade 0) to 52% (grade 1-2) and 8.7% (grade ≥3; P < .001; Figure 5B).

The R/R MM cohort comprised 191 patients treated with B-cell maturation antigen–directed CAR-Ts: 26% ide-cel; 57% ciltacabtagene autoleucel; and 17% orvacabtagene autoleucel (supplemental Table 12). The median age was 65 years, 97% of patients had received at least 3 lines of prior therapy, and the median prelymphodepletion platelet count was 143 × 10⁹/L.

In terms of grade distribution, 47% had no early T-ICAHT, 21% had grades 1 to 2, and 32% had grade ≥3 (Figure 5C). Early T-ICAHT grades were associated with OS, with 2-year OS estimates ranging from 80% (grade 0) to 62% (grade 1-2) and 37% (grade ≥3; P < .001; Figure 5D).

This combined MSK and National Cancer Institute (NCI) cohort comprised of 138 adult and 135 pediatric patients treated with CD19 CAR-Ts for B-ALL: 11% with tisa-cel, 4% with brexu-cel, and the rest with investigational CAR products such as CD19/28z (supplemental Table 13).²⁹ The median age was 19 years, 66% of patients had received at least 3 lines of prior therapy, and 45% were previously treated with an allogeneic hematopoietic cell transplantation. Of interest, we noted an association between higher early T-ICAHT grades and increased baseline bone marrow (BM) blast burden, with a median blast count of 53% in the patients who went on to develop grade 3 to 4 early T-ICAHT (supplemental Table 13).

The incidence of grade 3 to 4 early T-ICAHT was highest in B-ALL among all the disease cohorts in this study, at 51%, including 20% with grade 4 manifestations (Figure 5E). Of interest, a significantly higher proportion of adult patients with B-ALL had grade 4 early T-ICAHT than the pediatric data set (29% vs 11%; P < .001; supplemental Figure 3A-B). We again noted an association between early T-ICAHT severity and OS. The 2-year OS estimates ranged from 62% (grade 0) to 42% (grade 1-2) and 23% (grade ≥3; P < .001; Figure 5F). These survival differences, especially the inferior survival in patients with grade 3 to 4 early ICAHT, were consistently observed in both the pediatric and adult cohorts (supplemental Figure 3C-D).

This study highlights thrombocytopenia as a common and clinically relevant complication of CAR-T therapy for hematological malignancies. The proposed T-ICAHT grading system, based on the depth and duration of thrombocytopenia, effectively stratifies patients with different patterns of platelet recovery and identifies those with higher transfusion needs and poorer survival outcomes. Initially developed for patients with NHL treated with commercial CD19-targeted CAR-Ts, we further demonstrate that the T-ICAHT grading system has utility in independent cohorts of patients with NHL receiving investigational products, as well as in patients with MM and ALL. Overall, the T-ICAHT grading system serves as a valuable tool for describing platelet-specific hematological toxicity consistently and reproducibly, both in clinical practice and in the context of investigational therapies.

The applied thresholds for early (days 0-30) and late T-ICAHT (beyond day 30) were closely aligned with the existing neutropenia-based grading.⁹ Importantly, this enables seamless integration into a future unified ICAHT grading encompassing the most clinically relevant cytopenias of CAR-T therapy.⁹ Accordingly, this study has actively informed ongoing efforts by an ASTCT expert panel to revise major immune effector cell (IEC)-related toxicities. Regarding the specific cutoff values, it is acknowledged that any partition of continuous variables is inherently somewhat arbitrary. Similar to the Common Terminology Criteria for Adverse Events definitions,³⁰ cutoffs of 50 × 10⁹/L and 20 × 10⁹/L were applied. The 20 × 10⁹/L threshold, in particular, holds clinical significance because it is commonly used to indicate the need for platelet transfusion, especially in febrile patients or those in outpatient settings. Furthermore, these cutoffs have been traditionally used to describe platelet engraftment after hematopoietic cell transplantation and are already being recorded in major CAR-T registries, such as EBMT and the Center of International Blood and Marrow Transplantation. Moreover, cutoffs <20 × 10⁹/L may not reliably reflect actual platelet counts, because they depend on transfusion policies at individual centers.

Thrombocytopenia was relatively common in the NHL cohort, with ∼40% of patients experiencing platelet counts <50 × 10⁹/L and ∼20% falling below 20 × 10⁹/L, consistent with prior reports.^4,7,20,31,32 As expected, thrombocytopenia was more frequent in patients with ALL, followed by MM, with the lowest rates observed in NHL.⁷ Interestingly, platelet counts after CAR-T therapy exhibited a biphasic pattern: an initial decline likely linked to lymphodepletion, followed by a second drop starting around day 15 to 20, possibly related to the inflammatory sequelae of CAR-T expansion, which might explain the higher frequency of severe T-ICAHT observed in patients developing CRS (supplemental Table 3).³³ These patterns varied by T-ICAHT grade, with more severe T-ICAHT showing lower initial platelet counts, a more pronounced second drop, and delayed or even incomplete recovery.

Although this study focused on the clinical outcomes of post–CAR-T thrombocytopenia, it is important to note that T-ICAHT also captured the risk of other cytopenias. Its association with neutropenia is particularly interesting. Despite some correlation with N-ICAHT grading, significant discordance was observed. For instance, nearly 40% of patients with grade ≥3 T-ICAHT had grade ≤2 N-ICAHT. This underscores the importance of a platelet-specific toxicity definition and the need to report N-ICAHT and T-ICAHT as separate entities to adequately capture the full spectrum of post–CAR-T cytopenia. Additionally, it should be noted that long-term transfusion support for platelets and RBCs can also be observed with CAR-T therapy.³⁴ Notably, apart from the obvious observation of increased platelet transfusion support, most patients with high RBC transfusion burden were also captured as either grade 3 or 4 T-ICAHT.

The risk of severe early T-ICAHT was higher in patients with impaired functional status, elevated pre–CAR-T HEMATOTOX score, and those who received bridging therapy, whereas patients with FL had a lower risk. The HEMATOTOX score, which includes markers associated with hematopoietic reserve and baseline inflammation, has been previously validated as a predictor of cytopenias in patients with LBCL,²⁰ MCL,³² and MM.³⁵ Furthermore, a modified version substituting BM disease burden for serum ferritin was recently developed and validated for pediatric and adult patients with B-ALL.¹⁶ Taken together, these results expand the utility of the score to the risk discrimination for thrombocytopenia arising after CAR-T therapy, particularly because the association between severe T-ICAHT and the HEMATOTOX score was detected across several disease entities in this study (Figure 3; supplemental Tables 11 and 12).

An important finding of this study was the ability of T-ICAHT, both early and late, to associate with survival outcomes independently of established risk factors in patients with NHL treated with commercial CAR T cells. Furthermore, it stratified OS in an independent cohort of patients with NHL treated with an investigational product and in patients with MM and B-ALL. This result was somewhat unexpected, because severe bleeding due to thrombocytopenia is not a major contributor to mortality.³⁶ Although the precise mechanisms underlying the association between T-ICAHT and survival remain unclear, it can be hypothesized that platelet-specific toxicity serves as a reliable marker of BM reserve, which may itself reflect pre–CAR-T treatment intensity and disease burden. Furthermore, thrombocytopenia may serve as a readout of CAR-mediated inflammation and endothelial dysfunction with concomitant capillary leak.^28,37 Its potent prognostic role also provides context for the utility of the Endothelial Activation and Stress Index (EASIX) score in CAR-T recipients, which incorporates the platelet count in the denominator.^26,27,38

Some limitations of this study arise from its retrospective design, which included patients over an extended follow-up period. During this time, significant advances occurred in the field, including changes in the management of complications, earlier use of CAR-T therapy in the disease course, and availability of different products.³⁹ Nonetheless, the comprehensive real-world experience outlined herein, combined with the inclusion of a substantial number of patients with NHL treated with point-of-care products and the use of a fully automated computational pipeline, has highlighted the broad applicability of the T-ICAHT grading. The most salient clinical application of the grading lies in its capacity to realize transparent, reproducible, and standardized assessment of post–CAR-T thrombocytopenia. This will allow for comprehensive comparison of thrombocytopenia across different disease entities and CAR-T products. For example, our study revealed a numeric trend toward less pronounced thrombocytopenia with liso-cel relative to axi-cel in patients with R/R LBCL, as had been suggested in initial studies.⁴⁰ This may be a factor of both intensity of lymphodepletion and costimulatory signal used in the CAR-T construct, as well as additional patient and disease factors.

Taken together, the novel T-ICAHT grading system, which incorporates both the depth and duration of thrombocytopenia, provides a robust framework to capture and describe distinct patterns of platelet-specific hematological toxicity after CAR-T therapy for various hematological malignancies. Furthermore, it effectively stratifies survival outcomes. These findings support the integration of T-ICAHT into the broader ICAHT grading system to enhance the characterization of early and late post–CAR-T cytopenias.

K.R. received funding from the Else Kröner Forschungskolleg (EKFK) within the Munich Clinician Scientist Program, the Bruno and Helene Jöster Foundation, and the “CAR-T Control” translational group within the Bavarian Center for Cancer Research (number TLG-22). N.N.S. is supported by the Intramural Research Program, Center for Cancer Research, National Cancer Institute (NCI), and National Institutes of Health (NIH) Clinical Center (ZIA BC 011823). All authors from Memorial Sloan Kettering Cancer Center (MSKCC) were supported by a MSKCC Core grant (P30 CA008748) from the NIH/NCI. R.S. was supported by an NIH-NCI K-award (K08CA282987), Swim Across America, and Comedy vs Cancer.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Contribution: K.R., J.S., M.-A.P., and R.S. conceptualized and designed the study; K.R., J.S., M.S.N., H.H., A.A., O.B.-K., V.L.B., K.J.C., S.E., J.H.E., N.G., N.G.-A., M.G.-L., I.H., O.I., F.L.L., S.M., R.M., M.V.M., M.L.P., J.H.P., M.P., S. Raj, S. Rajeeve, G.S., M. Scordo, G.L.S., A.S., M. Subklewe, T.T., S.Z.U., O.V., Y.K.V., T.Z., N.N.S., M.-A.P., and R.S. conducted investigation and wrote, reviewed, and edited the manuscript; K.R., J.S., T.F., and R.S. contributed to formal analysis and data visualization; K.R., J.S., T.F., N.N.S., M.-A.P., and R.S. provided methodology; K.R., J.S., and R.S. wrote the original draft; and all authors read and approved the final manuscript.

Conflict-of-interest disclosure: K.R. reports research funding from Kite/Gilead; honoraria from Novartis and Kite/Gilead, Bristol Myers Squibb; consultancy fees from Kite/Gilead, Bristol Myers Squibb/Celgene, and CSL Behring; and travel support from Pierre Fabre and Kite/Gilead. J.S. reports honoraria from Pfizer, Merck Sharp & Dohme, Gilead, Novartis, Kite, Jazz, and Sanofi; and consultancy fees from Kite, Novartis, Jazz, Sanofi, Takeda, and MaaT Pharma. H.H. reports honoraria from Janssen and Karyopharm. V.L.B. reports honoraria from Amgen, Pfizer, Kite/Gilead, and Novartis; research funding from Celgene and Kite/Gilead; and consultancy/advisory fees from Novartis, Bristol Myers Squibb, and Takeda. K.J.C. has served as consultant for Novartis; board member for Turn Bio and PromiCell; and received research funding from Novartis, Cellectis, and Atara Bio. S.E. reports research funding from Merck Sharp & Dohme. M.G.-L. reports research funding from Sociedad Española de Hematología y Hemoterapia; and honoraria from Janssen. F.L.L. reports research funding from 2seventy bio, Allogene Therapeutics, Bristol Myers Squibb, Kite/Gilead, and Novartis; consultancy roles with A2 Biotherapeutics, Allogene Therapeutics, Amgen, bluebird bio, Bristol Myers Squibb, Calibr, Caribou Biosciences, Celgene, Cellular Biomedicine Group, Cowen, EcoR1 Capital, Emerging Therapy Solutions, Gerson Lehrman Group, GammaDelta Therapeutics, Gilead, Iovance, Janssen, Kite/Gilead, Legend Biotech, Novartis, Pfizer, Sana, Umoja, and Wugen; honoraria from the American Society of Hematology, Aptitude Health, BioPharma, Clinical Care Options Oncology, Communications CARE Education, iMedX, and the Society for Immunotherapy of Cancer; travel support from American Society of Hematology and Kite/Gilead; and patents and royalties for cellular immunotherapy from the Moffit Cancer Center. S.M. received consulting fees from EviCore, Optum, Bio Ascend, Janssen Oncology, Bristol Myers Squibb, AbbVie, ECor1, Galapagos, and Legend Biotech; Memorial Sloan Kettering Cancer Center receives research funding from the National Cancer Institute, Leukemia and Lymphoma Society, Janssen Oncology, Bristol Myers Squibb, Allogene Therapeutics, Fate Therapeutics, Caribou Biosciences, and Takeda Oncology for research led by S.M.; and received honoraria from OncLive, Physician Education Resource, MJH Life Sciences, and Plexus Communications. M.V.M. has equity and serves on the board of directors of 2seventy bio; receives research funding from Kite, Moderna, and Sobi; and has equity in A2Bio, AffyImmune Therapeutics, and Model T bio; and has served as a consultant for multiple companies developing cell therapies. M.L.P. has stock and other ownership interests in Seres Therapeutics; reports honoraria from Seres Therapeutics (I); consulting or advisory role in Seres Therapeutics (I), BeiGene, Synthekine, MustangBio, Cellectar, Bristol Myers Squibb, and Novartis; research funding from Seres Therapeutics; and patents, royalties, and other intellectual property with Intellectual Property Rights and Juno Intellectual Property Rights. M.P. has received fees for consulting or advisory role from Bristol Myers Squibb; research funding from Kite, Novartis, Bristol Myers Squibb, Janssen, and GlaxoSmithKline. J.H.P. received consulting fees from AffyImmune Therapeutics, Amgen, Autolus, Be Biopharma, BeiGene, Bristol Myers Squibb, Bright Pharmaceutical Services, Inc, Caribou Biosciences, Curocell, Galapagos, IN8bio, Kite, Medpace, Minerva Biotechnologies, Pfizer, Servier, Sobi, and Takeda; honoraria from OncLive, Physician Education Resource, and MJH Life Sciences; serves on scientific advisory boards of Allogene Therapeutics, Artiva Biotherapeutics, and Green Cross Biopharma; and received institutional research funding from Autolus, Genentech, Fate Therapeutics, Incyte, Servier, and Takeda. G.S. has received in the last 12 months financial compensation for participating in advisory boards from AbbVie, BeiGene, Bristol Myers Squibb, Genentech/Roche, Genmab, Incyte, Ipsen, Janssen, Kite/Gilead, Merck, Novartis, and Pfizer; and research support from AbbVie, Genentech, Genmab Janssen, Ipsen, and Nurix, which was managed by his institution. M. Scordo reports consultancy fees from McKinsey & Company, Angiocrine Bioscience, Inc, and Omeros Corporation; received research funding from Angiocrine Bioscience, Inc, Omeros Corporation, Amgen Inc, Bristol Myers Squibb, and Sanofi; served on ad hoc advisory boards for Kite/Gilead and Miltenyi Biotec; and received honoraria from i3 Health, Medscape, Cancer Network, Intellisphere LLC, Curio Science LLC, and IDEOlogy. G.L.S. has research funding to the institution from Janssen, Amgen, Bristol Myers Squibb, BeyondSpring, GPCR, and Recordati; and is on the data and safety monitoring board for Arcellx. M. Subklewe receives industry research support from Amgen, Bristol Myers Squibb/Celgene, Gilead, Janssen, Miltenyi Biotec, MorphoSys, Novartis, Roche, Seattle Genetics, and Takeda; and serves as a consultant/adviser to AvenCell, CDR-Life, Ichnos Sciences, Incyte Biosciences, Janssen, Molecular Partners, and Takeda; serves on the speakers’ bureau for Amgen, AstraZeneca, Bristol Myers Squibb/Celgene, Gilead, GlaxoSmithKline, Janssen, Novartis, Pfizer, Roche, and Takeda. S.Z.U. reports research funding (past 2 years) from AbbVie, Bristol Myers Squibb/Celgene, GlaxoSmithKline, Gilead, Gracell Biotechnologies, and Janssen; and consulting fees (past 2 years) from AbbVie, Bristol Myers Squibb/Celgene, Genentech, Gilead, GlaxoSmithKline, Janssen, Kite/Arcellx, Oricell Therapeutics, Pfizer, Regeneron, and Sanofi. Y.K.V. received a 1-time consulting fee from EastRx. N.N.S. receives research funding from Lentigen, Vor Bio, and Cargo Therapeutics; has attended advisory board meetings (no honoraria) for VOR, ImmunoACT, and Sobi; and receives royalties from Cargo Therapeutics. M.-A.P. reports honoraria from Adicet Bio, AlloVir, Caribou Biosciences, Celgene, Bristol Myers Squibb, Equilium, ExeVir, Incyte, Karyopharm, Kite/Gilead, Merck, Miltenyi Biotec, MorphoSys, Nektar Therapeutics, Novartis, Omeros Corporation, Orca Bio, Syncopation, VectivBio AG, and Vor Biopharma; serves on data and safety monitoring boards for Cidara Therapeutics, Medigene, and Sellas Life Sciences; and on the scientific advisory board of NexImmune; has ownership interests in NexImmune and Omeros Corporation; and has received institutional research support for clinical trials from Incyte, Kite/Gilead, Miltenyi Biotec, Nektar Therapeutics, and Novartis. R.S. reports honoraria from Sanofi and Incyte. The remaining authors declare no competing financial interests.

Correspondence: Roni Shouval, Adult Bone Marrow Transplantation and Cellular Therapy Services Memorial Sloan Kettering Cancer Center, 530 E74th St, New York, NY 10021; email: shouvalr@mskcc.org.