Improved outcomes with TKI maintenance after brexucabtagene autoleucel in Philadelphia chromosome ALL Authors: Tamer Othman, Gregory W Roloff, Katharine Miller, Amy Zhang, Katherine C Sutherland, Rawan Faramand, LaQuisa C Hill, Ibrahim N Muhsen, Nikeshan Jeyakumar, Abdullah Ladha, Josh Sasine, Chenyu Lin, Noam E Kopmar, Stephanie B Tsai, Timothy E O'Connor, Talal Hilal, Melhem M Solh, Caitlin Guzowski, Virginia Tan, Navneet S Majhail, Minoo Battiwalla, Shahbaz Malik, John Mathews, Paul Shaughnessy, Luke Mountjoy, Kaitlyn C Dykes, Aaron C Logan, Marc Schwartz, Matthew P Connor, Jiasheng Wang, Sumithira Vasu, Muthu Kumaran, Rasmus Hoeg, Sean I Tracy, Fevzi Yalniz, George Yaghmour, Vinod Pullarkat, Ryan D Cassaday, Jessica T Leonard, Bijal D Shah Published: 2026 ABSTRACT Key Points•In adult patients who achieve MRD− CR after brexu-cel, TKI maintenance is associated with lower incidence of relapse and improved PFS.•No differences in nonrelapse mortality were observed in patients receiving TKI maintenance vs no maintenance. Introduction Brexucabtagene autoleucel (brexu-cel) leads to high complete response (CR) rates in relapsed or refractory (R/R) B-cell acute lymphoblastic leukemia (B-ALL), including patients harboring the BCR::ABL1 fusion.1 However, relapse remains common, even after early achievement of measurable residual disease–negative (MRD−) response to brexu-cel.2 In an attempt to reduce subsequent relapse, consolidative allogeneic hematopoietic cell transplantation (HCT) is often pursued after CD19–chimeric antigen receptor (CD19-CAR) T-cell therapy,3,4 but this approach may be limited by patient suitability, transplant-related toxicity, nonrelapse mortality (NRM), and patient and clinician preferences.5 These considerations underscore the need for alternative strategies for remission consolidation after brexu-cel. The use of ABL1 tyrosine kinase inhibitors (TKIs) as maintenance after CD19-CAR T-cell therapy represents a rational approach for patients with Philadelphia chromosome–positive (Ph+) B-ALL. Previous studies have demonstrated that TKI maintenance after HCT in Ph+ ALL is both safe and beneficial, contributing to a reduced incidence of relapse and improved survival.6,7 However, the published experience with TKI maintenance after CAR T-cell therapy in Ph+ ALL has been limited to case reports and small case series.8, 9, 10 Determining the feasibility and use of TKI maintenance after CAR T-cell therapy is important, particularly in patients who are unable or unwilling to undergo consolidative HCT. Furthermore, given conflicting preclinical findings that TKIs may negatively affect CAR T-cell proliferation and/or persistence, studying this intervention in the clinical realm is warranted.11,12 Using data abstracted from 19 CAR T-cell centers across the United States, we evaluated the impact of maintenance TKI on relapse and survival outcomes after commercial brexu-cel for R/R Ph+ ALL. Methods We conducted a multicenter retrospective study using the Real-World Outcomes Collaborative for CAR T in ALL (ROCCA) consortium database.4 Each institution received institutional review board approval or exemption and transferred data to a centralized, Health Insurance Portability and Accountability Act-compliant electronic database at Stanford University. The study cohort included patients aged ≥18 years with a diagnosis of Ph+ B-ALL who received standard-of-care commercial brexu-cel between 2021 and 2025 and achieved measurable MRD− CR as best response after brexu-cel. In this study, MRD was determined by multicolor flow cytometry, as well as quantitative polymerase chain reaction (PCR) for BCR::ABL1 and/or immunoglobulin/T-cell receptor gene quantification by next-generation sequencing (NGS; supplemental Table 1). Patients who received HCT as remission consolidation at any time point after brexu-cel or any type of maintenance other than with a TKI were excluded. The cohort was stratified into 2 groups based on the receipt of TKI maintenance. Administration of TKI maintenance, including the decision to use maintenance, choice of agent, dosing, and duration, was not standardized across the ROCCA consortium centers and was at the discretion of the treating physician. All patients who received TKI maintenance were confirmed to be in MRD− CR before TKI initiation. Comparisons of baseline characteristics were made using χ2 or Fisher exact tests for categorical variables and the Wilcoxon rank-sum test for continuous variables. The primary end point was progression-free survival (PFS), defined as the time from brexu-cel infusion to either clinical relapse (defined as ≥5% bone marrow blasts, development of extramedullary or central nervous system disease, or disease detected by PCR or NGS for which therapy was altered) or death. Secondary end points included overall survival (OS), defined as the time from brexu-cel infusion to death, cumulative incidence of relapse, and NRM. Cox proportional hazards models were used to estimate hazard ratios (HRs) and 1-year survival probabilities for PFS and OS. Fine-Gray subdistribution hazards models were used to calculate HRs and cumulative incidence rates of relapse, with death as a competing risk, and NRM, with relapse as a competing risk. Both unadjusted and adjusted models were constructed, with TKI maintenance included as a time-dependent variable. The adjusted models controlled for key baseline differences, specifically T315I mutation. Statistical analyses were performed using SAS version 9.4 (Cary, NC). The institutional review board approved this study at each participating site. This study was conducted in accordance with the Declaration of Helsinki. Results Among 107 patients with Ph+ ALL treated with brexu-cel reported to the ROCCA database, 60 achieved CR and had adequate documentation of MRD− status. After excluding patients for consolidative HCT (n = 7) or non-TKI maintenance (n = 2), 51 patients formed the analytic cohort: 20 received TKI maintenance, and 31 received no maintenance (Figure 1). CONSORT diagram. The 2 cohorts were well balanced across patient and disease characteristics and previous treatments (Table 1). The median age in both cohorts was 53 years, and the median previous lines of therapy in both cohorts was 4. Most patients receiving TKI maintenance vs no TKI maintenance had low disease burden (<5% blasts) at apheresis (87% vs 85%), had undergone previous HCT (85% vs 74%), and had exposure to ponatinib before brexu-cel (70% vs 65%). Rates of maximum-grade cytokine release syndrome and immune effector cell–associated neurotoxicity syndrome did not differ across cohorts. Among patients receiving maintenance TKI, ponatinib (75%) and asciminib (15%) were the most common. The median time to TKI initiation after brexu-cel infusion was 91 days (interquartile range, 66-107). Table 1.Baseline characteristicsTKI maintenancen = 20No TKI maintenancen = 31P valueAge, median (range), y53 (31-72)53 (27-70).78Female10 (50)19 (61).43Ethnicity/race.045 Black1 (5)5 (16) Hispanic9 (45)4 (13) Non-Hispanic White10 (50)20 (65) Other0 (0)2 (6)Isoform (n = 47)∗.55 p19012 (63)20 (71) p2107 (37)8 (29)Additional cytogenetic abnormalities (n = 50)∗5 (25)8 (27)1.00IKZF1del (n = 20)∗5 (50)1 (10).14IKZF1plus (n = 20)∗†2 (20)0.47T315I ABL1 kinase domain mutation (n = 29)∗3 (21)9 (60).06Median no. of previous lines (range)4 (2-13)4 (2-9).71Previous blinatumomab10 (50)15 (48).91Previous alloHCT17 (85)23 (74).49Previous ponatinib14 (70)20 (65).69Bridging therapy16 (80)22 (71).47Disease status at preapheresis (n = 48)∗.78 MRD− CR5 (25)7 (26) MRD+ CR10 (50)11 (41) Active disease5 (25)9 (33)Preapheresis bone marrow disease burden (n = 42)∗1.00 <5% blasts13 (87)22 (85) 5%-49% blasts1 (7)1 (4) ≥50% blasts1 (7)3 (12)EMD or CNS disease at LD8 (40)7 (23).18Max CRS.73 None4 (20)5 (13) Grade 1-216 (80)24 (77) Grade 3-40 (0)2 (6)Max ICANS.94 None7 (35)12 (39) Grade 1-25 (25)8 (26) Grade 3-48 (40)11 (35)TKI used for maintenance-- Imatinib1 (5)-- Dasatinib1 (5)-- Ponatinib15 (75)-- Asciminib3 (15)--Time from brexu-cel to TKI, median (IQR), d91 (66, 107)----Cumulative exposure to TKI, median (range), d820 (26, 1186)--Median follow-up time, mo23.119.2--Numbers are shown as n (%) or median (range).CNS, central nervous system; CRS, cytokine release syndrome; EMD, extramedullary disease; ICANS, immune effector cell–associated neurotoxicity syndrome; IQR, interquartile range; LD, lymphodepletion; Max, maximum.∗Percentages were calculated based on those who were tested for these characteristics.†Defined as IKZF1del and deletion in CDKN2A, CDKN2B, PAX5, or PAR1 with no ERG deletion.13 Baseline characteristics Numbers are shown as n (%) or median (range). CNS, central nervous system; CRS, cytokine release syndrome; EMD, extramedullary disease; ICANS, immune effector cell–associated neurotoxicity syndrome; IQR, interquartile range; LD, lymphodepletion; Max, maximum. Percentages were calculated based on those who were tested for these characteristics. Defined as IKZF1del and deletion in CDKN2A, CDKN2B, PAX5, or PAR1 with no ERG deletion.13 The median follow-up time from brexu-cel infusion was 23.1 months and 19.2 months for the maintenance and no maintenance cohorts, respectively. Thirteen relapses occurred (1 in the maintenance cohort and 12 in the no maintenance cohort). The estimated 1-year cumulative incidence of relapse was 5% (95% confidence interval [CI], 1-27) in the maintenance cohort vs 32% (95% CI, 19-54) in the non-maintenance cohort (HR, 0.12; 95% CI, 0.02-0.88; P = .037; Figure 2A). Eleven deaths occurred (2 in the maintenance cohort and 9 in the no maintenance cohort; supplemental Table 2). The estimated 1-year PFS was 91% (95% CI, 80-100) in the maintenance cohort vs 62% (95% CI, 47-81) in the no maintenance cohort (HR 0.19; 95% CI, 0.04-0.85; P = .029; Figure 1B). The estimated 1-year OS was excellent for both groups with a nonsignificant trend toward improvement in the maintenance cohort: 96% (95% CI, 89-100) vs 88% (95% CI, 78-99), respectively (HR, 0.34; 95% CI, 0.07-1.62; P = .18; Figure 1C). Finally, the estimated 1-year cumulative incidence of NRM did not differ between cohorts (HR, 0.83; 95% CI, 0.12-5.87; P = .85) and was estimated at 5% (95% CI, 1-26) in the maintenance cohort vs 6% (95% CI, 2-21) in the no maintenance cohort (Figure 1D). Univariate logistic regression analysis showed that the presence of a T315I mutation or extramedullary or central nervous system leukemia were not associated with an increased risk of relapse, with odds ratios of 2.33 (95% CI, 0.41-13.17; P = .34) and 0.65 (95% CI, 0.15-2.80; P = .56), respectively. Of 13 relapses observed in the entire cohort, 12 were tested for T315I, and 4 (33%) were confirmed to have a T315I mutation. To adjust for any impact of T315I mutation on the effect of TKI maintenance, we performed a multivariate analysis, including only the 29 patients who had T315I mutation analysis. After adjusting for T315I mutation status, TKI maintenance was associated with a trend toward improved PFS with an adjusted hazard ratio of 0.21 (95% CI, 0.04-1.10; P = .065) without an increase in NRM (HR, 0.51; 95% CI, 0.02-16.25; P = .71) (Table 2). In contrast to the findings with TKI maintenance, we found no statistically significant difference in PFS (HR, 0.87; 95% CI, 0.23-3.26; P = .83), OS (HR, 1.51; 95% CI, 0.30-7.72; P = .62), relapse (HR, 1.35; 95% CI, 0.32-5.80; P = .68), or NRM (HR, 0.55; 95% CI, 0.02-19.96; P = .75) based on T315I mutation status. Outcomes between TKI and no TKI maintenance groups. (A-D) Comparison of cumulative incidence of relapse (A), PFS (B), OS (C), and NRM (D) between the TKI maintenance and the no TKI maintenance groups. Table 2.Unadjusted and adjusted association of TKI maintenance with survival outcomesModelPFSOSRelapseNRMHR (95% CI)P valueHR (95% CI)P valueHR (95% CI)P valueHR (95% CI)P valueUnadjusted0.19 (0.04-0.85).0290.34 (0.07-1.62).180.12 (0.02-0.88).0370.83 (0.12-5.87).85Adjusted∗0.21 (0.04-1.10).0650.53 (0.09-3.20).490.18 (0.02-1.39).100.51 (0.02-16.25).71HR is compared with the no maintenance reference group.∗Model adjusted for T315I mutation. Unadjusted and adjusted association of TKI maintenance with survival outcomes HR is compared with the no maintenance reference group. Model adjusted for T315I mutation. Discussion The use of TKI, delivered with the intention to maintain deep remissions after CAR T-cell therapy for Ph+ ALL, remains controversial and nonstandardized, with limited data to guide clinical decision-making. To our knowledge, this represents the largest study evaluating the value of TKI maintenance among patients with Ph+ ALL who achieve MRD− CR after CAR T-cell therapy. We found that when used as prophylactic maintenance after brexu-cel, TKI therapy is associated with a lower incidence of subsequent relapse and improved PFS, without increasing NRM. Although the efficacy of TKI maintenance for enhancing CAR T-cell–mediated remission may be attributed to direct BCR::ABL1 inhibition, additional mechanisms may contribute.14 TKIs have been shown to have immunomodulatory potential, with evidence that second-generation ABL1 TKIs decrease regulatory T cells and enhance effector and memory CD8+ T-cell populations.15 Moreover, preclinical data from a murine model suggest that dasatinib may inhibit CAR T-cell differentiation and exhaustion, thereby promoting antileukemic efficacy.16 This is countered by other preclinical evidence demonstrating that dasatinib may be inhibitory to CAR T-cell function when administered on a continuous basis but may promote CAR T-cell persistence when provided on an on-off pulsed schedule.12 Unfortunately, in our data set, CAR T-cell kinetics, as measured by direct CAR detection or B-cell aplasia as a surrogate, were not routinely obtained across sites, limiting analysis of any potential clinical effect of TKI maintenance on CAR T-cell persistence or function. We acknowledge that our study has several limitations, including its retrospective, observational design, relatively small sample size, and short follow-up duration. As a result of this, clinically meaningful imbalances may persist despite nonsignificant P values. In particular, differences in disease burden at infusion and the possibility that patients selected for maintenance represented a more favorable post–brexu-cel subset. We intended to study as homogeneous a population of commercially treated patients as possible and thus excluded patients who did not achieve MRD− CR after brexu-cel, those with early deaths precluding use of TKI, patients receiving other forms of post-CAR consolidation or maintenance, and patients in whom MRD− disease could not be confirmed before TKI initiation. To reduce immortal time bias, we used a time-varying covariate approach with all cumulative incidence and survival analyses; however, we recognize that additional biases may exist regarding patient selection and heterogeneity of real-world treatment practices. Furthermore, MRD assessment via NGS measuring immunoglobulin/T-cell receptor expression has been shown to be more prognostic than BCR::ABL1 quantitative PCR in Ph+ ALL.17,18 NGS MRD data were not available for all patients in this cohort, which limits interpretation. Moreover, multivariable analyses were limited due to sample size. Although we were able to perform a small adjusted Cox model including T315I mutation status, T315I resistance mutation testing was not performed in 43% of patients included in this analysis, and thus, only 29 patients were included, limiting statistical power. Finally, although data on CAR T-cell expansion and persistence in the 2 cohorts would have provided useful insight on the impact of TKI maintenance on CAR T-cell kinetics, none of the patients in the TKI maintenance cohort underwent expansion and persistence testing. Brexu-cel is characterized by relatively short in vivo persistence, previously reported as a median of 53.5 days.19 Therefore, interventions that negatively affect CAR T-cell persistence may have greater consequences for longer-lasting products, such as those incorporating a 4-1BB costimulatory domain. Whether the benefit of TKI maintenance observed with brexu-cel extends to other commercially available products, such as tisagenlecleucel and obecabtagene autoleucel, remains unknown. In summary, in this analysis of adult patients with R/R Ph+ ALL who achieved MRD− CR after brexu-cel, our results suggest that TKI maintenance is associated with improved PFS, without an increase in NRM. Although our study supports offering prophylactic TKI maintenance after brexu-cel, due to the limitations discussed, these findings are hypothesis-generating and require prospective validation. Conflict-of-interest disclosure: G.W.R. reports advisory board fees from Kite/Gilead, Autolus, and Rigel. R.F. reports consulting or advisory role fees from Autolus and Kite/Gilead; and honoraria from Sanofi. A.L. reports research funding from Kite/Gilead (Institutional), Kadmon/Sanofi (Institutional), Amgen (Institutional), Autolus (Institutional), and Talaris (Institutional). J.S. reports research funding from Miltenyi Biotec. C.L. reports consulting or advisory role fees from Autolus, Rigel, and ADC Therapeutics. S.B.T. reports consulting or advisory role fees from Gamida Cell, Syndax, Bristol Myers Squibb (BMS), and Sanofi. T.H. reports research funding from BeiGene (Institutional) and BMS (Institutional). M.M.S. reports consulting or advisory role fees from Kite and Autolus; and honoraria from BMS. L.C.H. reports advisory board fees from March Biosciences, Autolus, Sanofi, and Arcellx; and speaker’s bureau fees from Kite/Gilead. P.S. reports consulting or advisory role fees from Autolus, BMS, and Kite/Gilead; honoraria from Autolus, BMS, Kite/Gilead, and Sanofi; and speakers' bureau fees from Sanofi and BMS. A.C.L. reports consulting or advisory role fees from Adaptive Biotechnologies, Amgen, BMS, Pfizer, Sanofi, Kite/Gilead, Orca Bio, Servier, and Takeda. M.K. reports consulting or advisory role fees from Kite/Gilead; and speaker’s bureau fees from Geron. G.Y. reports consulting or advisory role fees from Alexion, Novartis, Omeros, and Pfizer; and speakers' bureau fees from GSK, Rigel, Servier, Jazz Pharmaceuticals, BMS, Kite, Sobi, Incyte, Syndax, Blueprint Medicines, and Secura Bio. V.P. reports advisory committees/speakers’ bureau fees and honoraria from AbbVie, Amgen, Genentech, Jazz Pharmaceuticals, Novartis, Pfizer, Rigel, and Servier. R.D.C. reports consulting or advisory role fees from Autolus, Amgen, Jazz Pharmaceuticals, Kite/Gilead, Pfizer, and PeproMene Bio; honoraria from Autolus, Amgen, Jazz Pharmaceuticals, Kite/Gilead, and Pfizer; research funding from Amgen, Kite/Gilead, Incyte, Merck, Pfizer, Servier, and Vanda Pharmaceuticals; stock and other ownership interests in Seagen (Spouse); and employment in Seagen (Spouse). J.T.L. reports consulting or advisory role fees from Pfizer, Adaptive Biotechnologies, Takeda, Kite/Gilead, Autolus, and Amgen; and research funding from AbbVie. B.D.S. reports consulting or advisory role fees from Amgen, Kite/Gilead, Lilly and Loxo Oncology, Autolus, Adaptive Biotechnologies, Jazz Pharmaceuticals, BMS, Novartis, Syndax, AstraZeneca, Pfizer, and BeiGene; research funding from AstraZeneca, Autolus, Kite/Gilead, Precision BioSciences/Imugene, and Servier; travel, accommodations, and expenses from Amgen; and data safety monitoring board fees from BeiGene, PeproMene Bio, and Precision BioSciences/Imugene. N.V.F. reports consulting or advisory role fees from Kite/Gilead, Pfizer, and Autolus. L.S.M. reports consulting or advisory role fees from Amgen, Astellas Pharma, Kite/Gilead, Pfizer, Vor Bio, CARGO Therapeutics, and Autolus; honoraria from UpToDate; research funding from Adaptive Biotechnologies, Astellas Pharma, Jasper Therapeutics, Kite/Gilead, and BMS; and stock and other ownership interests in Corvus Pharmaceuticals. I.A. reports consulting or advisory role fees from Amgen, AstraZeneca, Autolus, Jazz Pharmaceuticals, Kite/Gilead, Pfizer, Sobi, Syndax, Takeda, and Wugen; honoraria from Amgen; and research funding from AbbVie, MacroGenics, and Novartis. The remaining authors declare no competing financial interests. Acknowledgments Authorship Contribution: T.O., L.S.M., and I.A. wrote and edited the manuscript; and all authors collected data and approved the manuscript. Footnotes Data are available from the corresponding author, Ibrahim Aldoss (ialdoss@coh.org), on request. The full-text version of this article contains a data supplement. Supplementary Material Supplemental Tables References Associated Data Supplementary Materials Supplemental Tables