Highlights
- The 2026 EBMT/CTIWP registry study by Dachy and colleagues provides the largest real-world analysis to date of fludarabine dose variation during lymphodepletion in large B-cell lymphoma (LBCL) treated with CD19 CAR-T therapy.
- Within the tisagenlecleucel cohort, higher cumulative fludarabine exposure (82.6-120 mg/m²) did not improve disease control and was associated with inferior overall survival versus standard dosing (67.5-82.5 mg/m²).
- Across product groups, axicabtagene ciloleucel showed superior progression-free and overall survival compared with tisagenlecleucel, but at the cost of more immune effector cell-associated neurotoxicity syndrome (ICANS).
- The study reinforces a central clinical principle: lymphodepletion is necessary for CAR-T efficacy, but dose escalation of fludarabine is not a simple surrogate for better CAR-T expansion or superior lymphoma control.
Background
Relapsed or refractory LBCL remains a high-risk clinical setting despite progress with anti-CD19 CAR T-cell therapy. In the last decade, three autologous CD19-directed products have shaped the field: axicabtagene ciloleucel (axi-cel), tisagenlecleucel (tisa-cel), and lisocabtagene maraleucel (liso-cel). These products differ in co-stimulatory domain, manufacturing characteristics, expansion kinetics, and toxicity profile, but all rely on preinfusion lymphodepleting chemotherapy to create an immunologic niche that permits CAR-T expansion and persistence.
The standard lymphodepletion backbone in commercial CD19 CAR-T therapy is fludarabine plus cyclophosphamide. The biologic rationale is well established: lymphodepletion depletes endogenous lymphocytes and regulatory populations, reduces cytokine sinks, increases homeostatic cytokines such as IL-7 and IL-15, and may diminish host-versus-graft immune rejection of infused CAR-T cells. Yet, the optimal intensity of lymphodepletion remains uncertain in daily practice, especially in older, heavily pretreated, renally vulnerable patients with aggressive lymphoma.
This question has practical urgency because real-world CAR-T programs often individualize fludarabine dose for age, renal function, frailty, center preference, or concern regarding prior therapies. At the same time, insufficient lymphodepletion could compromise CAR-T expansion, whereas excessive fludarabine might worsen infectious risk, prolonged cytopenias, neurotoxicity susceptibility, or nonrelapse mortality without improving tumor control. The new EBMT study directly addresses this tension in a large registry-based LBCL population.
Key Content
1. The new EBMT study: design, cohort, and main findings
Dachy et al. analyzed 1,498 patients with LBCL treated from 2019 to 2023 in the EBMT registry: 549 received tisa-cel and 949 received axi-cel. The core hypothesis was whether variation in fludarabine dose during lymphodepletion altered outcomes. A notable operational finding was that fludarabine dosing varied substantially for tisa-cel, whereas axi-cel was largely administered with standard-dose lymphodepletion. This natural experiment created an opportunity to compare standard- versus higher-dose fludarabine within the tisa-cel cohort and then benchmark these groups against axi-cel.
In the tisa-cel-only analysis, higher fludarabine dosing (82.6-120 mg/m² cumulative) was associated with worse overall survival than standard-dose fludarabine (67.5-82.5 mg/m²), with a hazard ratio of 1.29 (95% CI, 1.02-1.64; p=0.036). This is a clinically important negative result: more fludarabine did not translate into superior benefit and may instead identify or contribute to harm.
When the investigators compared three groups, relapse incidence remained higher after tisa-cel than after axi-cel, and dose escalation of fludarabine did not rescue this efficacy gap. Relative to axi-cel, relapse risk was elevated with both standard-dose tisa-cel (HR 1.69; 95% CI, 1.39-2.06; p<0.001) and high-dose tisa-cel (HR 1.45; 95% CI, 1.09-1.94; p=0.012). Axi-cel also yielded superior progression-free survival and overall survival. However, this greater efficacy came with a familiar trade-off: more ICANS.
From a translational perspective, the study suggests that product-specific biology dominates over modest variation in lymphodepletion intensity, at least within the dosing ranges observed. In other words, attempting to pharmacologically compensate for lower efficacy with more fludarabine appears ineffective in LBCL treated with tisa-cel.
2. Why lymphodepletion matters in CAR-T therapy
The concept that preparative chemotherapy is integral rather than incidental to adoptive cell therapy predates modern commercial CAR-T products. Preclinical and early clinical studies showed that host lymphodepletion augments transferred T-cell expansion by removing competing endogenous lymphocytes and increasing availability of homeostatic cytokines. In CAR-T programs, fludarabine and cyclophosphamide emerged as the dominant platform because they are immunologically potent, operationally feasible, and clinically familiar.
Mechanistically, fludarabine is not merely cytotoxic chemotherapy. It contributes to profound lymphocyte depletion, impairs host immune rejection of infused cells, and may influence antigen-presenting cell and cytokine networks that shape peak CAR expansion. Cyclophosphamide adds both cytoreduction and immunomodulation, including depletion of regulatory T-cell populations and facilitation of cytokine release patterns conducive to adoptive T-cell engraftment. The combination has repeatedly outperformed weaker or non-fludarabine-containing approaches in early cellular therapy experience.
Yet there is no biologic law stating that progressively more fludarabine will necessarily improve outcomes. Beyond a threshold, additional immunosuppression may intensify infections, marrow suppression, mucosal injury, immune dysregulation, and hospitalization burden. Moreover, CAR-T efficacy in LBCL depends on many variables that lymphodepletion cannot fully overcome: baseline tumor burden, systemic inflammation, T-cell fitness, manufacturing attributes, bridging therapy exposure, histologic aggressiveness, and CAR construct design.
3. Product-specific efficacy signals in LBCL: placing the EBMT results in context
The EBMT analysis should be interpreted against pivotal and real-world product data in LBCL.
Axi-cel, built on a CD28 co-stimulatory domain, showed strong activity in ZUMA-1 and set an early benchmark in refractory LBCL. Long-term follow-up confirmed durable remissions in a subset of patients, alongside substantial rates of cytokine release syndrome (CRS) and neurotoxicity. Tisa-cel, using a 4-1BB co-stimulatory domain, demonstrated efficacy in JULIET with a generally different toxicity profile, especially lower high-grade neurotoxicity compared with axi-cel in cross-trial impressions, although such comparisons are intrinsically indirect. Liso-cel later added another 4-1BB product with favorable efficacy and comparatively manageable toxicity in TRANSCEND NHL 001.
Cross-trial comparison must be cautious, but several real-world series and registry analyses have repeatedly suggested somewhat higher early disease control with axi-cel than with tisa-cel in aggressive LBCL, with the anticipated cost of more ICANS. The EBMT study is therefore not wholly surprising in its product-level findings. What is new is the demonstration that increasing fludarabine exposure in the tisa-cel setting does not abolish this difference.
This point matters clinically because practitioners sometimes wonder whether inferior outcomes in one product group may partly reflect “undertreated” lymphodepletion. The present registry data argue against that explanation as a dominant driver.
4. Pivotal evidence for CAR-T products in LBCL
In ZUMA-1, axi-cel produced high response rates in refractory LBCL and established commercial CD19 CAR-T as a transformative option for patients with previously dismal prognosis. Later updates showed a plateau in progression-free survival consistent with durable remission in a meaningful minority of patients.
In JULIET, tisa-cel also produced durable responses in relapsed/refractory LBCL, but operational differences, broader real-world eligibility patterns, and distinct product kinetics complicate direct juxtaposition with axi-cel. The key point is that both products are active, but their efficacy-to-toxicity balance differs.
TRANSCEND NHL 001 extended the field with liso-cel, emphasizing controlled composition and a lower observed incidence of severe neurologic toxicity in many reports. Although liso-cel is not part of the EBMT fludarabine-dose comparison presented here, its existence reminds clinicians that the relationship between product design and toxicity may be more consequential than modest variation in preparative chemotherapy.
Second-line randomized trials further underscored product-specific differences. ZUMA-7 established axi-cel superiority over standard salvage chemotherapy followed by autologous transplantation in eligible second-line LBCL. TRANSFORM similarly favored liso-cel. By contrast, BELINDA did not show superiority for tisa-cel over standard care in that setting. These data are not directly about lymphodepletion dose, but they support the broader interpretation that CAR construct, manufacturing vein-to-vein time, and disease control during bridging are major determinants of outcome that may not be correctable through intensified fludarabine.
5. How should clinicians interpret the apparent harm of higher fludarabine dose?
The observed association between higher-dose fludarabine and worse overall survival in tisa-cel recipients could reflect causation, confounding, or both.
First, higher fludarabine exposure may truly cause more toxicity. Fludarabine is associated with prolonged lymphopenia, opportunistic infections, delayed immune reconstitution, and cumulative marrow suppression, especially in older or renally impaired patients. In a population already prone to cytopenias from prior chemoimmunotherapy, stem-cell collection attempts, bridging therapy, and marrow involvement, this can plausibly worsen survival independently of relapse.
Second, higher dosing may mark adverse baseline characteristics not fully captured in retrospective data. Centers may have chosen intensified lymphodepletion for patients with aggressive disease, bulky tumor burden, or concern for poor CAR expansion. If so, dose escalation could be a surrogate for physician perception of high-risk biology. Registry methods can adjust for measured confounders, but residual confounding is difficult to eliminate.
Third, the biologic effect of more fludarabine may be nonlinear. Once a sufficient lymphodepletion threshold is reached, further dose increases may not materially improve CAR expansion. Above that threshold, toxicity may accumulate faster than efficacy. This threshold model is consistent with the EBMT findings.
Fourth, renal function deserves particular attention. Fludarabine pharmacokinetics are sensitive to renal clearance. A nominally “higher” prescribed dose may effectively represent much greater systemic exposure in patients with reduced kidney function, potentially amplifying neurotoxicity and infectious complications. Most registry analyses cannot fully capture exposure-response relationships without pharmacokinetic data.
6. The ICANS trade-off: why axi-cel may still outperform despite more neurotoxicity
One of the most clinically relevant messages in the EBMT study is the persistence of the classic CAR-T trade-off: axi-cel delivered superior efficacy but more ICANS. This aligns with accumulated experience that CD28-based products expand more rapidly and robustly, often translating into higher early antitumor activity but also more inflammatory toxicity.
Neurotoxicity after CAR-T is multifactorial and linked to endothelial activation, cytokine flux, blood-brain barrier dysfunction, macrophage activation, and immune-cell trafficking into the central nervous system. Lymphodepletion can modulate this environment but is unlikely to be the sole determinant. Thus, increasing fludarabine in tisa-cel recipients would not be expected to reproduce the same expansion kinetics or efficacy profile as axi-cel, because the intracellular signaling architecture of the CAR differs fundamentally.
Clinically, these findings support individualized product selection. Patients with highly aggressive LBCL who need maximal disease control may derive greater benefit from axi-cel when eligible and when experienced neurotoxicity management is available. Patients at elevated neurologic risk, frailty risk, or with center-specific logistics may still be appropriate for tisa-cel or other products, but the expectation should not be that intensified fludarabine compensates for product differences.
7. Registry evidence versus randomized evidence: strengths and limitations of the EBMT study
The major strength of the EBMT analysis is scale. A sample size approaching 1,500 patients across routine European practice offers a powerful view of how lymphodepletion is actually delivered outside highly selected trials. It captures center heterogeneity, broader comorbidity, and clinically relevant dose variation that would be difficult to study prospectively.
The study is also hypothesis-generating in a highly actionable way. Unlike many translational correlates, fludarabine dose is immediately modifiable. Therefore, a strong registry signal against escalation carries practical weight.
However, the study remains retrospective and nonrandomized. Important variables may be incompletely recorded, including renal dose adjustment rationale, exact tumor burden, inflammatory markers, timing and type of bridging therapy, infection history, CNS risk, and center-specific supportive care pathways. Product allocation itself is not random and may reflect disease tempo, manufacturing timelines, and national reimbursement patterns. The observed superiority of axi-cel versus tisa-cel, while consistent with prior literature, should still be interpreted in light of these potential biases.
An additional limitation is that cumulative body-surface-area dosing is an imperfect proxy for true exposure. Pharmacokinetic studies of fludarabine could prove more informative than dose bands alone, particularly in older patients and those with renal dysfunction. Future studies should examine area-under-the-curve-based dosing, immune reconstitution markers, cytokine profiles, and CAR expansion kinetics together.
8. Translational and practical implications
Several practice implications emerge.
First, standard fludarabine/cyclophosphamide lymphodepletion remains the preferred default in LBCL CAR-T therapy unless there is a clear medical reason for adjustment. The EBMT analysis does not support routine fludarabine escalation in tisa-cel-treated patients.
Second, product choice should remain primarily driven by disease biology, logistics, comorbidity profile, and toxicity tolerance rather than an assumption that preparative chemotherapy can neutralize intrinsic product differences.
Third, pharmacology matters. Dose personalization should focus less on empiric escalation and more on avoidance of unintended overexposure, especially in renal impairment. This is an underdeveloped area in CAR-T medicine and a fertile target for prospective study.
Fourth, endpoints beyond response are crucial. If intensified fludarabine does not improve relapse but worsens survival, then harms such as infection, prolonged cytopenias, hospitalization, and late nonrelapse events may be mediating the effect. These deserve systematic capture in future registry and translational cohorts.
Fifth, the study highlights an unmet need for prospective optimization of lymphodepletion. Questions include whether pharmacokinetically guided fludarabine dosing, biomarker-stratified conditioning, or alternative lymphodepletion backbones might improve therapeutic index in selected patients.
Expert Commentary
The EBMT report is important not because it changes the biology of CAR-T therapy, but because it clarifies where therapeutic leverage probably does and does not reside. Clinicians have long recognized that lymphodepletion is indispensable. The tempting corollary has been that more lymphodepletion might mean better CAR-T performance. Dachy et al. challenge that assumption in the largest LBCL registry analysis yet.
The study’s negative result is clinically constructive. It suggests that once adequate lymphodepletion is delivered, the dominant drivers of outcome are product-intrinsic signaling, patient fitness, tumor biology, and treatment logistics. This is consistent with broader CAR-T experience: rapid manufacturing, disease control during bridging, lower inflammatory burden at infusion, and healthier starting T cells all matter profoundly. No simple increase in fludarabine is likely to override those factors.
From a policy and systems perspective, the analysis also supports standardization. Wide center-level variation in lymphodepletion may not be benign. Harmonized protocols, renal function-adjusted dosing frameworks, and prospective toxicity capture should become part of quality improvement in CAR-T programs.
The most interesting scientific question raised by the study is not whether high-dose fludarabine is ineffective, but why. Is the problem excess exposure in biologically vulnerable patients? Is there a ceiling effect for host immune depletion? Or does intensified conditioning simply identify clinicians’ attempts to rescue poor-risk disease? Answering these questions will require integrated datasets that combine dose, pharmacokinetics, cytokines, marrow reserve, CAR expansion, and competing-risk outcomes.
Until such data are available, the most defensible interpretation is pragmatic: do not escalate fludarabine routinely in tisa-cel-treated LBCL hoping to improve efficacy. If greater disease control is needed and the patient is an appropriate candidate, product selection and timing likely matter more than more intense fludarabine.
Conclusion
The 2026 EBMT/CTIWP retrospective study provides the clearest real-world evidence so far that higher fludarabine dose during lymphodepletion does not improve outcomes for tisa-cel-treated LBCL and may be associated with inferior overall survival. In parallel, the study reinforces a broader pattern already visible across pivotal trials and registries: axi-cel generally delivers stronger efficacy than tisa-cel in LBCL, at the expense of more ICANS.
For clinicians, the immediate message is straightforward. Lymphodepletion is essential, but more is not necessarily better. Standard-dose fludarabine/cyclophosphamide remains the evidence-aligned approach in most patients, and product-specific biology cannot be easily overcome by conditioning intensification. Future progress will likely come from better patient selection, pharmacologically informed dosing, optimized bridging, earlier referral, and biomarker-guided strategies rather than empiric escalation of fludarabine.
References
- Dachy G, Mooyaart JE, Gabellier L, et al. Impact of fludarabine dosage on outcomes in large B-cell lymphoma patients treated with CAR T-cell therapy: a retrospective study of the CTIWP of the EBMT. Bone Marrow Transplant. 2026. PMID: 42251173.
- Locke FL, Neelapu SS, Bartlett NL, et al. Phase 1 Results of ZUMA-1: A Multicenter Study of KTE-C19 Anti-CD19 CAR T Cell Therapy in Refractory Aggressive Lymphoma. Mol Ther. 2017;25(1):285-295. PMID: 27939001.
- Neelapu SS, Locke FL, Bartlett NL, et al. Axicabtagene Ciloleucel CAR T-Cell Therapy in Refractory Large B-Cell Lymphoma. N Engl J Med. 2017;377(26):2531-2544. PMID: 29226797.
- Schuster SJ, Bishop MR, Tam CS, et al. Tisagenlecleucel in Adult Relapsed or Refractory Diffuse Large B-Cell Lymphoma. N Engl J Med. 2019;380(1):45-56. PMID: 30501490.
- Abramson JS, Palomba ML, Gordon LI, et al. Lisocabtagene maraleucel for patients with relapsed or refractory large B-cell lymphomas (TRANSCEND NHL 001): a multicentre seamless design study. Lancet. 2020;396(10254):839-852. PMID: 32888407.
- Locke FL, Miklos DB, Jacobson CA, et al. Axicabtagene Ciloleucel as Second-Line Therapy for Large B-Cell Lymphoma. N Engl J Med. 2022;386(7):640-654. PMID: 34891224.
- Kamdar M, Solomon SR, Arnason JE, et al. Lisocabtagene maraleucel versus standard of care with salvage chemotherapy followed by autologous stem-cell transplantation as second-line treatment in patients with relapsed or refractory large B-cell lymphoma (TRANSFORM): results from an interim analysis of an open-label, randomised, phase 3 trial. Lancet. 2022;399(10343):2294-2308. PMID: 35717989.
- Bishop MR, Dickinson M, Purtill D, et al. Second-Line Tisagenlecleucel or Standard Care in Aggressive B-Cell Lymphoma. N Engl J Med. 2022;386(7):629-639. PMID: 34904798.
- Nastoupil LJ, Jain MD, Feng L, et al. Standard-of-Care Axicabtagene Ciloleucel for Relapsed or Refractory Large B-Cell Lymphoma: Results From the US Lymphoma CAR T Consortium. J Clin Oncol. 2020;38(27):3119-3128. PMID: 32501297.
- Turtle CJ, Hanafi LA, Berger C, et al. CD19 CAR-T cells of defined CD4+:CD8+ composition in adult B cell ALL patients. J Clin Invest. 2016;126(6):2123-2138. PMID: 27111235.