Clonal Hematopoiesis Is Associated with Increased Risk of Severe Neurotoxicity in Axicabtagene Ciloleucel Therapy of Large B-Cell Lymphoma

Adoptive T-cell transfer therapy with chimeric antigen receptor (CAR) T cells represents a breakthrough in the treatment of hematologic malignancies (1–3). Three CD19-CAR T-cell products have received approval from regulatory medical agencies and have been introduced into clinical practice for relapsed and refractory large B-cell malignancies (r/r LBCL; tisagenlecleucel, axicabtagene ciloleucel, and Iisocabtagene maraleucel; refs. 1–3). Although durable responses have been observed in 30% to 40% of r/r LBCL patients treated with CAR T-cell therapy (1–3), it is associated with significant systemic inflammatory toxicities, such as cytokine release syndrome (CRS) and immune effector-cell-associated neurotoxicity syndrome (ICANS), that are occasionally fatal (4). Treatment-related toxicities with severe grade ≥3 CRS and/or ICANS occur in 10% to 31% of patients receiving CAR T products (1–3). Although there has been remarkable progress in the understanding and clinical management of CAR T-related toxicities (5), a significant knowledge gap exists in the mechanisms and host factors impacting these toxicities.

Clonal hematopoiesis (CH) is a clonally expanded population of hematopoietic stem cells bearing somatic gene mutations (6). CH is recognized as a driver of systemic inflammation (7) and is associated with an increased risk of therapy-related myeloid neoplasms (t-MN) after chemotherapy (8, 9). Murine studies suggest that the knockout of CH-associated genes (Dnmt3a or Tet2) can contribute to a dysregulated inflammatory microenvironment by altering T-cell function (10, 11). Furthermore, clinical evidence indicates an emerging role of CH in accelerating graft versus host disease (GvHD) after allogeneic stem cell transplantation (12). As anti-CD19 CAR T-cell therapy, a highly effective therapy for LBCL and other lymphoid malignancies (13), is associated with systemic inflammatory toxicities, and given CH's role in driving systemic inflammation, we hypothesized that CH influences the incidence and severity of CAR T-cell therapy toxicities. This study aimed to identify the clinical impact of CH in r/r LBCL patients undergoing CAR T-cell therapy.

A total of 114 r/r LBCL patients at two different institutions, MD Anderson Cancer Center (MDACC, Houston, TX, n = 99) and Moffitt Cancer Center (Tampa, FL, n = 15), whose peripheral blood (PB) buffy coat samples were available for CH analysis, were studied. The patient characteristics of the study cohort are listed in Table 1. Of the 114 patients with r/r LBCL, 105 were treated with axicabtagene ciloleucel and 9 received tisagenlecleucel. The median age for the entire cohort was 63 years (range, 29–87 years) and patients received a median of 3 lines of systemic therapy prior to CAR T-cell therapy. The histologic diagnosis was subclassified into DLBCL/high-grade B-cell lymphoma (n = 91), transformed follicular lymphoma (n = 21), and primary mediastinal lymphoma (n = 2).

CH was detected in the pretreatment samples of 42 of the 114 (36.8%) patients. The complete list of genes and variants is provided in Supplementary Table S1. Blood parameters on day -5 prior to lymphodepleting chemotherapy, including serum inflammatory markers, such as ferritin and C-reactive protein, were not significantly different between patients with and without CH (Table 1; Supplementary Fig. S1A–S1C). The most frequently mutated genes were PPM1D (19/114, 16.7%), TP53 (13/114, 11.4%), DNMT3A (7/114, 6.1%), TET2 (6/114, 5.2%), and ASXL1 (4/114, 3.5%; Fig. 1; Supplementary Fig. S2). A total of 72 CH variants were detected in 42 patients with the median variant allele frequency (VAF) of CH of 5.8% (range, 2.1%–49.5%; Supplementary Fig. S3) and 19 variants in 15 patients were present at a VAF greater than 10%. In 30 (71.4%) patients, a single gene mutation was detected as CH, and 12 (28.6%) patients carried two or more gene mutations. The high proportion of patients having CH mutations in DNA damage pathway genes (PPM1D and TP53) was notable in this cohort (Supplementary Table S1) and is likely associated with prior exposure to chemotherapies (14–17). Among the 12 patients with more than one mutation, the most frequent combination was TP53 and PPM1D (n = 8, 44.4%) mutations.

The median duration of follow-up among survivors in our cohort was 14.9 (range, 1.2–30.5) months. The best overall response rate (ORR) and complete response (CR) for the whole cohort were 78.5% (84/107) and 56.1% (60/107), respectively. The rates of CR and ORR were not significantly different between patients with CH and without CH (CR: 55.0% vs. 56.7%, P = 1.00, ORR: 85.0% vs. 74.6%, P = 0.23; Fig. 2A). The median progression-free survival (PFS) and overall survival (OS) for the whole cohort were 4.8 and 15.7 months, respectively (Supplementary Fig. S4). We did not observe any significant differences in PFS and OS between patients with CH and those without CH (Fig. 2B; Supplementary Fig. S5; Supplementary Tables S2 and S3).

We also analyzed the impact of CH on CAR T-associated toxicities. A total of 39 (92.9%) and 65 (90.3%) patients developed CRS with all grades in the CH and no-CH cohorts, respectively (P = 0.743). A total of 24 (57.1%) and 37 (51.4%) patients developed ICANS with all grades in the CH and no-CH cohorts, respectively (P = 0.566). As we observed no differences in the incidence of all grades CRS or ICANS between the two cohorts, we next analyzed the incidence of severe toxicities (grades ≥3). There were 7 (6.1%) and 37 (32.5%) patients who had grade ≥3 CRS and grade ≥3 ICANS, respectively, in the entire population. Although the overall incidence of grade ≥3 CRS was low in our cohort (6.1%), the incidence was numerically higher, but not statistically significant, in patients with CH (9.5%, 4/42) compared with in patients without CH (4.2%, 3/72; P = 0.42; Fig. 2C). The rate of grade ≥3 ICANS was significantly higher in patients with CH, at 45.2% (19/42), compared with 25.0% (18/72 patients) in patients without CH (P = 0.038; Fig. 2C; Supplementary Table S4). On a multivariate analysis, the presence of CH was the only covariate significantly associated with an increased risk of grade ≥3 ICANS (odds ratio = 2.47; 95% CI, 1.02–6.02, P = 0.046; Supplementary Table S5). The percentage of patients requiring tocilizumab and corticosteroids for management of CRS and ICANS was comparatively higher in the CH group, although not statistically significant. In patients with CH, 64.3% (27/42) and 52.4% (22/42) required tocilizumab and corticosteroids, respectively, compared with 55.6% (40/72, P = 0.43) and 43.1% (31/72, P = 0.43) of patients in the no-CH group, respectively (Supplementary Fig. S6). Additionally, to make our cohort more homogenous, we analyzed a subset of 105 patients who received axicabtagene ciloleucel as the CAR T-cell construct. Among these 105 patients, the incidence of grade ≥3 ICANS was higher in patients with CH (46.3%, 19/41) compared with in patients without CH (26.5%, 17/64, P = 0.037). The incidence of grade ≥3 CRS was similar in the CH cohort (9.7%, 4/41) and the no-CH cohort (4.7%, 3/64, P = 0.3; Supplementary Table S6).

We further analyzed the survival and toxicity outcomes in patients whose CH mutations have been associated with inflammation in the literature, namely, DNMT3A, TET2, and ASXL1 (DTA mutations). In patients harboring DTA CH mutations, the incidence of grade ≥2 ICANS [70.5% (12/17) vs. 41.7% (30/72), P = 0.06] or ≥3 ICANS [58.9% (10/17) vs. 25% (18/72), P = 0.02] was significantly higher than in patients with no CH mutations (Table 2). Similarly, we saw a trend of increased grade ≥3 CRS in patients with DTA CH mutations compared with in patients with no-CH mutations [17.7% (3/17) vs. 4.2% (3/72), P = 0.08]. However, we did not find any difference in response rates between patients with DTA CH mutations and without CH mutations, as shown in Table 2. We compared the DTA CH cohort versus the no-CH cohort among the 105 patients who received axicabtagene ciloleucel. Similar to the results in the whole cohort, the incidence of grade ≥3 ICANS among the patients who received axicabtagene ciloleucel and had DTA CH mutations was significantly higher (58.9%, 10/17 vs. 26.5%, 17/64, P = 0.01), and there was a trend toward increased severe-grade CRS (17.6%, 3/11 vs. 4.7%, 3/64, P = 0.06) when compared with patients who received axicabtagene ciloleucel and had no CH mutations (Supplementary Table S7).

In patients with available samples, we also analyzed inflammatory cytokines in plasma at serial timepoints from day 0 until 2 weeks after CAR T infusion (n = 43). There was a trend of higher median plasma IL6 at day 0 in CH patients (1.12 pg/mL) compared with no-CH patients (0.62 pg/mL) (P = 0.058; Supplementary Table S8). However, no differences were seen in other inflammatory cytokines. Also, we did not observe statistically significant differences in peak plasma concentration of any inflammatory cytokine between the CH and no-CH patients (Supplementary Table S9).

We assessed for cytopenia at leukapheresis and at day 90 after CAR T infusion in both the CH and no-CH cohorts. There were no differences in hemoglobin, platelet counts, and absolute lymphocyte and neutrophil counts at the time of leukapheresis or at day 90 after CAR T infusion (Supplementary Table S10). We also compared the incidence of therapyrelated myeloid neoplasms (t-MN) after CAR T-cell therapies. Seven patients developed t-MN following CAR T-cell therapy; five (5/42, 11.9%) had baseline CH and two (2/72, 2.8%) did not. At 24 months, the estimated cumulative incidence rates of t-MN after CAR T-cell therapy were 19% (95% CI, 5.5%–38.7%) and 4.2% (95% CI, 0.3%–18.4%) for patients with and without CH, respectively (P = 0.028; Fig. 2D). The clinical history and mutation analysis of the 5 patients with CH who subsequently developed t-MN are presented in Supplementary Table S11. Mutational analyses were available in only three patients, and among these, mutations were shared between CH and t-MN in two patients. The VAF of CH mutations on the diagnostic bone marrow samples corresponded to the blasts burden in these two patients. Information about these mutations is presented in Supplementary Table S11.

In this cohort of heavily pretreated LBCL patients, we found that CH is associated with increased severe immune-mediated toxicities, particularly ICANS, following CAR T-cell therapy. Also, we found that CH did not affect survival outcomes or responses following CAR T-cell therapy. These findings add to the growing body of evidence linking CH with systemic inflammation in multiple clinical contexts such as atherosclerosis (18), GvHD (12, 19), and infection (20).

The incidence of CH in our cohort was approximately 40% and was similar to the reported incidence of CH (48%) in a mixed population of patients with lymphoma and myeloma undergoing CAR T-cell therapy at Dana-Farber Cancer Institute (DFCI; ref. 15). These incidences are higher compared with those observed in LBCL patients undergoing autologous stem cell transplant (ASCT, CH incidence of 25%–30%; refs. 14, 17). As a majority of the patients undergoing CAR T were previously treated with ASCT, it is likely that there is a stepwise increase in the incidence of therapy-related CH with iterative exposures to chemotherapies. Also, although almost all CH studies have used PB for mutation analysis, there are no previous studies that compare the detectability of mutations between PB and bone marrow. CH can likely be detected more frequently in bone marrow, given that early hematopoietic stem cells are the source of CH. What was also notable in this cohort, as well as in other heavily treated cohorts, is the preponderance of CH with DNA damage response pathway genes, such as PPM1D and TP53 mutations, which often cooccurred in the same patient. Although it is difficult to dissect the clonal relationship of these two cooccurring mutations, our previous study using single-cell analysis indicated a mutually exclusive relationship of the two mutations at the cellular level (21).

In their study, the DFCI group reported an increase in CAR T-associated toxicities in the CH cohort, as well as higher CR rates (15). There was a statistically significant increase in grade ≥2 CRS (77.8% with CH vs. 45.9% without CH, P = 0.042), albeit only in patients with age <60 years. However, in the overall LBCL cohort, there was no difference in the incidence of grade ≥2 CRS between patients with CH and patients without CH. Moreover, the incidence of grade ≥2 ICANS was comparatively much higher in CH patients (60% vs. 43%, P = 0.06), although not statistically significant. Congruent to our study, these findings suggest that the presence of CH mutations could be a potential modifier of the pathophysiology of ICANS and associated with a higher severity of ICANS. Each CH mutation is biologically different, and CH-harboring myeloid cells might spread in the tumor microenvironment differently (22), leading to unique interactions with CAR T cells and resulting in disparate toxicity outcomes. In fact, most of the high-grade toxicities in our cohort were driven by DNMT3A/TET2/ASXL1 mutations. Grade ≥3 CRS and grade ≥3 ICANS rates with these three mutations were 17.6% (3/17) and ∼60% (10/17), respectively. In the future, studies with larger cohorts will be able to not only tease out the impact of individual mutations in CAR T-related toxicities and outcomes, but also help in resolving the discordant results seen between our study and others (15, 23).

Although anti-IL6 therapy has been shown to mitigate the development of severe CRS, interventions to prevent the development of severe ICANS are under study. Consistent with this practice and the notion that CH is associated with systemic inflammation, we observed a statistically significant association between CH and severe ICANS. CH has the potential to influence outcomes and toxicity in CAR T-cell therapy through multiple mechanisms. In the tumor microenvironment, the severity of toxicities could be influenced by cross talk between CH-mutant myeloid cells, tumor cells, and CAR T cells. This cross talk could also influence the activation of bystander immune cells and lymphocytes in the tumor milieu, potentially leading to more inflammation and toxicities. Moreover, these CH mutants are associated with differential metabolism requirements and can produce microchanges in the metabolic signatures associated with tumor stroma (24). In our cohort, we did not see major differences in inflammatory cytokines, either at baseline or at peak post-CAR T infusion, between the CH and no-CH cohorts. There was a trend of elevated baseline IL6 in the CH cohort in comparison with the no-CH cohort. Indeed, CH is associated with elevated IL6 in various murine models and in human samples (25, 26). Therefore, high baseline IL6 in patients with CH could be driven by the presence of CH, but more patient samples are needed to confirm this. Furthermore, the cytokine repertoire we examined did not include IL1, which has been shown to be well-associated with CH, especially in cases driven by the TET2 mutation (27). Moreover, IL1 is strongly associated with ICANS pathophysiology (5), and therefore, it will be important to explore the contribution of IL1 in CH-associated CAR T toxicities.

In agreement with prior reports (8, 9, 14), we observed an increased rate of t-MNs in patients with CH receiving lymphodepleting chemotherapy prior to CAR T-cell infusion, which portends poor outcomes. It is quite possible that through a longer follow-up and with a larger cohort, we might see poor outcomes in the CH cohort that is driven by a higher incidence of t-MNs, as seen in transplant settings (14). Taken together, our results suggest that further studies are needed to elucidate the biological mechanisms by which CH influences immune-mediated toxicities associated with CAR T-cell therapy. Our findings are largely applicable to axicabtagene ciloleucel constructs and more studies are needed to fully understand the association in tisagenlecleucel therapy. Understanding the mechanisms by which CH influences toxicities may lead to novel intervention strategies to prevent high-grade CRS and ICANS after CAR T-cell therapy.

Cryopreserved PB buffy coat samples from patients with r/r LBCL receiving standard-of-care axicabtagene ciloleucel or tisagenlecleucel CAR T-cell therapy collected at any time within 21 days preceding lymphodepleting chemotherapy were analyzed for CH detection. Only patients with r/r LBCL who received axicabtagene ciloleucel or tisagenlecleucel were included in this study, whereas those with other malignancies were excluded. Buffy coats were stored in liquid nitrogen after collection and were thawed immediately before processing and DNA extraction. The MDACC cohort (n = 99) consisted of consecutive LBCL patients who underwent anti-CD19 standard-of-care CAR T-cell therapy between October 2018 and June 2020 and whose frozen buffy coats were available in the Lymphoma Tissue Bank. Similarly, the Moffitt cohort included standard-of-care consecutive CAR T patients whose PB was available for analysis. Lymphodepleting chemotherapy is standardized at both institutions among standard-of-care CAR T-cell therapy with a dosing regimen of cyclophosphamide 500 mg/m² and fludarabine 30 mg/m² intravenously on the fifth, fourth and third days before infusion of axicabtagene ciloleucel CAR T-cell therapy. For the tisagenlecleucel construct, the dosing is slightly lower, with fludarabine at 25 mg/m² and cyclophosphamide 250 mg/m² daily on the fifth, fourth, and third days. The clinical management of CRS and ICANS was followed as per the MDACC CAR T-cell therapy-associated TOXicity (CARTOX) management algorithm (28) and did not include prophylactic tocilizumab in both centers. All patients provided written informed consent through an institutional review board-approved protocol either at The University of Texas MD Anderson Cancer Center (N = 99) or at the Moffitt Cancer Center (N = 15). The study was completed in accordance with the Declaration of Helsinki.

For the MDACC cohort, DNA was extracted and purified using a QIAamp DNA Mini Kit (Qiagen), according to the manufacturers’ protocol. The pretreatment buffy coat samples were sequenced using a SureSelect custom panel of 300 genes (Agilent Technologies) that covers genes recurrently mutated in CH and in hematologic malignancies. The complete descriptions of the sequencing methods and bioinformatics pipelines to identify high-confidence somatic single-nucleotide variants and indels from targeted capture DNA sequencing have been described in detail previously (12). For the Moffitt cohort, DNA was extracted from PB for library preparation using a custom 76-gene hybrid-capture panel with unique molecular barcodes. This panel, combined with deep next-generation sequencing, optimally captures whole exons of all commonly mutated CH genes, including single-nucleotide mutations and small indels down to allele frequencies ≥1% (29). Library preparation and sequencing were conducted in collaboration with the Moffitt Molecular Genomics Core Facility. DNA libraries were generated using the SureSelect^XTHS kit (Agilent) and sequenced on a NextSeq 2000 sequencer (Illumina) per manufacturers’ recommendations, with a goal of achieving an average coverage >1,000×. CH mutations were identified following our previously published bioinformatics pipeline (9, 29). Briefly, sequencing reads were aligned to the human genome (GRCh38) using the BWA-MEM algorithm (30). Somatic variant calling was performed using Genome Analysis Toolkit best practices (31). Variants were filtered and annotated using BCFtools (32). Mutations that occurred in greater than one-third of the samples were considered sequencing artifacts and removed. Additional quality control filters included read depth >50, strand odds ratio <3, TLOD >6.3, observation of each variant more than once on the forward and reverse reads, and masking of the repetitive regions of the genome as defined by the DUST algorithm (33). Germline variants were removed using publicly available reference populations (i.e., variants observed in noncancer populations at a frequency >0.005 and with a VAF >0.45 were removed; ref. 34). To filter for likely functional somatic variants, only mutations or indels located in exonic regions were considered; synonymous mutations and nonsynonymous mutations that had not been annotated in the Catalogue of Somatic Mutations in Cancer (COSMIC) database (35) were excluded. The remaining variants with a VAF > 0.02 were considered CH mutations. For all samples, we used a minimum VAF cutoff of 2% for CH mutations, in accordance with a prior report (6).

Available frozen plasma samples from the patients were obtained at different time points from day 0 to day 14 after infusion and cytokines were measured using multiplex assays on a Meso-Scale Discovery platform (36). The cytokines that were measured included IL2, IL4, IL5, IL16, IL10, IL13, IL17A, granulocyte—macrophage colony-stimulating factor (GM-CSF), tumor necrosis factor-alpha (TNF-a), and interferon-gamma (IFN-g).

Categorical covariates were summarized by frequencies and percentages, and continuous covariates were summarized by means, standard deviations, medians, and ranges. Box-and-whisker plots were also used to summarize continuous variables. Comparisons between cohorts were performed using Fisher exact tests for categorical variables and Wilcoxon rank-sum tests for continuous variables. A multivariable logistic regression model was fitted to evaluate associations between CH and ICANS adjusting covariates of interest. Unadjusted survival distributions were estimated by the Kaplan–Meier method and comparisons were made with the log-rank test. Univariate Cox proportional hazards regression models were used to evaluate the associations between survival outcomes and the covariate of interest. The outcome variable of t-MNs was analyzed using competing risk models, where the competing risk was death. Gray's test was used for comparisons of t-MNs between cohorts. PFS was defined as the time from the date of CAR T infusion to the progression of disease or death or last follow-up (whichever occurred earlier). OS was defined as the time from the date of CAR T-cell infusion to death or last follow-up (whichever occurred earlier). A P value of <0.05 (two-tailed) was considered statistically significant. Statistical analyses were conducted using R 3.6.1 and GraphPad PRISM 9 software.

The data have been deposited with links to BioProject accession number PRJNA832976 in the NCBI BioProject database (http://www.ncbi.nlm.nih.gov/bioproject/832976).

D.M. Swoboda reports other support from Thermo Fisher outside the submitted work. U. Greenbaum reports grants from American Physicians for Israel outside the submitted work. P. Strati reports other support from Roche Genentech, TG Therapeutics, ADC Therapeutics, Hutchinson Medipharma, AstraZeneca Acerta, and ALX Oncology, grants from Lymphoma Research Foundation and NIH outside the submitted work. R. Nair reports personal fees from ScienciaCME during the conduct of the study; personal fees from Incyte outside the submitted work; and ScienciaCME Speaker/Preceptorship. S.P. Iyer reports grants from Seattle Genetics, Merck, Rhizen, Trillium, Innate, Crispr, Acrotech, Legend, and Ono, grants and other support from Curio Science, Targeted Oncology outside the submitted work. R. Steiner reports grants from Seagen, BMS, GSK, and Rafael Pharmaceutical outside the submitted work. N. Jain reports grants, personal fees, and nonfinancial support from Genentech, AbbVie, AstraZeneca, Cellectis, Fate Therapeutics, and Precision Biosciences outside the submitted work. L. Nastoupil reports grants and personal fees from Gilead/Kite, BMS, Novartis, Janssen, and Takeda and grants from Caribou Biosciences during the conduct of the study; grants and personal fees from Genentech and Epizyme, personal fees from ADC Therapeutics, Morphosys, and MEI grants from IGM Biosciences outside the submitted work. S. Loghavi reports grants from Astellas and Amgen outside the submitted work. M. Wang reports other support from AstraZeneca, Acerta Pharma, BeiGene, CSTone, BioInvent, DTRM Biopharma (Cayman) Limited, Celgene, Epizyme, Genentech, Genmab, Innocare, Janssen, Juno Therapeutics, Kite Pharma, Lilly, Loxo Oncology, Miltenyi Biomedicine GmbH, Molecular Templates, Oncternal, Pharmacyclics, VelosBio, and Vincerx during the conduct of the study; other support from AstraZeneca, Acerta Pharma, Anticancer Association, CStone, BeiGene, BioInvent, Celgene, DTRM Biopharma (Cayman) Limited, Epizyme, Genentech, Genmab, InnoCare, Janssen, Juno Therapeutics, Kite Pharma, Lilly, Loxo Oncology, Miltenyi Biomedicine GmbH, Molecular Templates, Oncternal, Pharmacyclics, VelosBio, Vincerx, Anticancer Association, BGICS, CAHON, Chinese Medical Association, Dava Oncology, Eastern Virginia Medical School, Hebei Cancer Prevention Federation, Imedex, Leukemia and Lymphoma Society, LLC TS Oncology, Medscape, Moffit Cancer Center, Mumbai Hematology Group, OMI, OncLive, Physicians Education Resources, Practice Point Communications, and The Frist Afflicted Hospital of Zhejiang University outside the submitted work. J.R. Westin reports personal fees from Kite/Gilead, Novartis, BMS, Morphosys/Incyte, Astra Zeneca, Genentech, ADC Therapeutics, Umoja, MonteRosa, Iksuda, Merck, and personal fees from Calithera and other support from Kymera outside the submitted work. M.R. Green reports grants from Kite/Gilead, Sanofi, Allogene, other support from Tessa Therapeutics, Monte Rosa Therapeutics, and Daiichi Sankyo outside the submitted work. D.A. Sallman reports other support from AbbVie, Aprea, Bluebird Bio, BMS, Incyte, Intellia, Janssen, Magenta, Servier, Shattuck Labs, Gilead, Syros, and Novartis outside the submitted work. M.L. Davila reports grants from Kite/Gilead, Novartis, and Precigen outside the submitted work. F.L. Locke reports grants from Leukemia and Lymphoma Society and NIH/NCI, personal fees from Allogene, BlueBird Bio, Calibr, Cellular BioMedicine Group, GammaDelta Therapeutics, Iovance, Janssen, Legend, Wugen, Sana, Takeda, and Umoja, personal fees and other support from BMS/Celgene, Kite Pharma, Novartis, other support from Cero during the conduct of the study; personal fees from Amgen, Cowen, EcoR1, Emerging Therapy Solutions, GLG, Aptitude Health, ASH, BioPharma Communications, CARE Education, Clinical Care Options, Imedex, and SITC outside the submitted work; in addition, F.L. Locke has a patent for US010414810B2 issued, a patent for 62/939,727 pending, a patent for 62/892,292 pending, a patent for 62/879,534 pending, a patent for PCT/US2021/01906 pending, a patent for 63/178,131 pending, a patent for 63/178,183 pending, and a patent for 63/178,317 pending. G. Garcia-Manero reports grants from Amphivena, grants and other support from Helsinn, Novartis, AbbVie, grants and other support from Bristol Myers Squibb, Astex, and Genentech, grants from Onconova, H3 Biomedicine, Merck, Curis, Janssen, Forty Seven, and Aprea outside the submitted work. P. Kebriaei reports personal fees from Kite Pharmaceuticals, Pfizer, and Jazz, and other support from Amgen and Ziopharm outside the submitted work. C.R. Flowers reports personal fees from BeiGene, BioAscend, Bristol Myers Squibb, Celgene, Curio Sciences, Denovo Biopharma, Epizyme/Incyte, Foresight Diagnostics, Genentech/Roche, Genmab, MEI Pharmaceuticals, MorphoSys AG, Pharmacyclics/Janssen, SeaGen and grants from 4D, AbbVie, Acerta, Adaptimmune, Allogene, Amgen, Bayer, Celgene, Cellectis, EMD, Gilead, Genentech/Roche, Guardant, Iovance, Janssen Pharmaceutical, Kite, Morphosys, Nektar, Novartis, Pfizer, Pharmacyclics, Sanofi, Takeda, TG Therapeutics, Xencor, Ziopharm, Burroughs Wellcome Fund, Eastern Cooperative Oncology Group, NCI, V Foundation, Cancer Prevention and Research Institute of Texas: CPRIT Scholar in Cancer Research outside the submitted work. M.D. Jain reports grants and personal fees from Kite/Gilead, grants from Incyte, and personal fees from BMS, Takeda, and Novartis during the conduct of the study. N. Gillis reports grants from NCI and the University of South Florida College of Medicine during the conduct of the study. S.S. Neelapu reports grants from Kite/Gilead, BMS, Cellectis, Poseida, Allogene, Unum Therapeutics, Precision Biosciences, Adicet Bio, personal fees from Kite/Gilead, Merck, Novartis, Sellas Life Sciences, Athenex, Allogene, Incyte, Adicet Bio, Calibr, BMS, Bluebird Bio, Sana Biotechnology, and other support from Takeda Pharmaceuticals outside the submitted work; in addition, S.S. Neelapu has a patent for cell therapy pending. K. Takahashi reports grants from NCI, American Society of Hematology, and Dresner Foundation during the conduct of the study; personal fees from Symbio Pharmaceuticals, GSK, Novartis, Dava Oncology, Mission Bio, BMS/Celgene, and Agios outside the submitted work. No disclosures were reported by the other authors.

N.Y. Saini: Conceptualization, data curation, formal analysis, visualization, methodology, writing–original draft, project administration, writing–review and editing. D.M. Swoboda: Writing–review and editing. U. Greenbaum: Writing–review and editing. J. Ma: Formal analysis, writing–review and editing. R.D. Patel: writing–review and editing. K. Devashish: Writing–review and editing. K. Das: Writing–review and editing. M.R. Tanner: Writing–review and editing. P. Strati: Writing–review and editing. R. Nair: Writing–review and editing. L. Fayad: Writing–review and editing. S. Ahmed: Writing–review and editing. H. Lee: Writing–review and editing. S.P. Iyer: Writing–review and editing. R. Steiner: Writing–review and editing. N. Jain: Writing–review and editing. L. Nastoupil: Writing–review and editing. S. Loghavi: Writing–review and editing. G. Tang: Writing–review and editing. R.L. Bassett: Writing–review and editing. P. Jain: Writing–review and editing. M. Wang: Writing–review and editing. J.R. Westin: Writing–review and editing. M.R. Green: Writing–review and editing. D.A. Sallman: Writing–review and editing. E. Padron: Writing–review and editing. M.L. Davila: Writing–review and editing. F.L. Locke: Writing–review and editing. R.E. Champlin: Writing–review and editing. G. Garcia-Manero: Writing–review and editing. E.J. Shpall: Writing–review and editing. P. Kebriaei: Writing–review and editing. C.R. Flowers: Writing–review and editing. M.D. Jain: Writing–review and editing. F. Wang: Writing–review and editing. A.P. Futreal: Writing–review and editing. N. Gillis: Writing–review and editing. S.S. Neelapu: Conceptualization, formal analysis, investigation, methodology, writing–review and editing. K. Takahashi: Conceptualization, formal analysis, methodology, project administration, writing–review and editing.

This work was supported in part by the University of Texas MD Anderson Cancer Center B-cell Lymphoma Moonshot (S.S. Neelapu), AML and MDS Moonshot (K. Takahashi), Sabin Family Fellow Award (K. Takahashi), American Society of Hematology Scholar Award (K. Takahashi), Lyda Hill Foundation (A.P. Futreal), Physician Scientist Program at MD Anderson (K. Takahashi), NIH/NCI R01 CA237291 (K. Takahashi), and NCI Cancer Center Support Grant to the University of Texas MD Anderson Cancer Center (P30 CA016672) and by the Molecular Genomics Core Facility and the Bioinformatics and Biostatistics Shared Resource at the H. Lee Moffitt Cancer Center and Research Institute, an NCI-designated Comprehensive Cancer Center (P30-CA076292).