Bendamustine lymphodepletion is a well‐tolerated alternative to fludarabine and cyclophosphamide lymphodepletion for axicabtagene ciloleucel therapy for aggressive B‐cell lymphoma

Fludarabine/cyclophosphamide (Flu/Cy) is established for lymphodepletion (LD) prior to standard‐of‐care CAR T‐cell therapy for lymphoma. There is ongoing need to test alternative LD regimens to preserve efficacy, improve safety, and address challenges including the recent national fludarabine shortage. We retrospectively evaluated outcomes among patients with relapsed/refractory aggressive B‐cell lymphoma who received bendamustine (n = 27) or Flu/Cy (n = 42) LD before axicabtagene ciloleucel (axi‐cel) at our institution. The median change in absolute lymphocyte count from pre‐LD to time of axi‐cel infusion was −0.6×109/L in bendamustine cohort and −0.7×109/L in Flu/Cy cohort. The best overall response/complete response rates were 77.8% (95% CI: 57.7%–91.4%)/48.1% (95% CI: 28.7%–68.1%) among bendamustine cohort and 81.0% (95% CI: 65.9%–91.4%)/50.0% (95% CI: 34.2%–65.8%) among Flu/Cy cohort. Six‐month progression‐free survival were 43.8% (95% CI: 24.7%–61.3%) and 55.6% (95% CI: 39.0%–69.3%) in bendamustine and Flu/Cy cohorts, while 6‐month overall survival were 81.5% (95% CI: 61.1%–91.8%) and 90.4% (95% CI: 76.4%–96.3%), respectively. Relative to Flu/Cy‐treated patients, bendamustine‐treated patients did not show an increase in hazards associated with experiencing progression/relapse/death (aHR:1.4 [95% CI: 0.7–2.8]; p = .32) or death (aHR:1.6 [95% CI: 0.5–5.6]; p = .46), after adjusting for baseline number of prior therapies and refractory disease. Any grade/grade ≥3 CRS were observed in 89%/3.7% and 86%/4.8% among bendamustine and Flu/Cy cohorts, while any grade ICANS/grade ≥3 ICANS were observed in 30%/19% and 55%/31% respectively. While more Flu/Cy‐treated patients experienced grade ≥3 neutropenia compared with bendamustine‐treated patients (100% vs. 68%), grade ≥3 infectious complications were comparable (24% vs. 19% respectively). More patients received bendamustine LD and axi‐cel as outpatient than Flu/Cy cohort, without increased toxicities and with shorter median inpatient stays. In conclusion, we observed comparable efficacy and lower any grade ICANS among patients receiving bendamustine relative to Flu/Cy LD, followed by axi‐cel.

infectious complications were comparable (24% vs. 19% respectively).More patients received bendamustine LD and axi-cel as outpatient than Flu/Cy cohort, without increased toxicities and with shorter median inpatient stays.In conclusion, we observed comparable efficacy and lower any grade ICANS among patients receiving bendamustine relative to Flu/Cy LD, followed by axi-cel.

| INTRODUCTION
Chimeric antigen receptor (CAR) T-cell therapies have changed the outcomes of patients with relapsed/refractory (R/R) aggressive B-cell lymphoma (aBCL).The use of lymphodepleting chemotherapy preceding the CAR T-cell infusion, commonly incorporating fludarabine and cyclophosphamide (Flu/Cy), is important for priming the immune environment for CAR T-cell expansion, persistence, and effector T-cell function. 1Fludarabine has been associated with improved CAR T-cell expansion and better clinical outcomes when combined with cyclophosphamide, 2,3 but can lead to neurotoxicity and delayed immune reconstitution. 4e ongoing national shortage of fludarabine 5 has necessitated an urgent need for alternative safe and effective lymphodepletion (LD) regimens for standard-of-care (SOC) CAR T-cell products at our and other CAR T centers.Bendamustine was an attractive choice given its anti-lymphoma activity, potent lymphodepleting effect, and good tolerability. 6The use of bendamustine-based LD regimens has been limited to SOC tisagenlecleucel (tisa-cel) and anti-CD30 CAR T-cell therapy on trial. 7th the long-term follow-up data from the JULIET trial, increased progression-free survival (PFS) was observed for Flu/Cy compared with bendamustine LD; however, these outcomes were not adjusted for patient characteristics (e.g., bendamustine resistance) confounding LD selection. 8In contrast, a retrospective multicenter study showed equivalent efficacy of bendamustine and Flu/Cy LD before tisa-cel infusion and lower incidence of hematologic toxicity and infection with bendamustine. 6In Hodgkin lymphoma treated with anti-CD30 CAR T-cells, fludarabine with bendamustine promoted a favorable cytokine profile (increased interleukin [IL]-7 and IL-15) compared with Flu/Cy or bendamustine alone, leading to higher antitumor activity and longer anti-CD30 CAR T-cell persistence. 7More evidence is required to establish whether bendamustine can be safely and effectively applied to other CAR T-cell products with different design and manufacturing process since expansion kinetics and persistence differ.While there are recent preliminary findings in a small number of patients of similar response rates and comparable axi-cel expansion with the use of bendamustine versus fludarabine LD, 9 more real-world data are required.
In this retrospective study, we report the safety and efficacy outcomes of utilizing bendamustine-based LD prior to axi-cel, in comparison with a consecutive case series of standard Flu/Cy LD among patients with R/R aBCL in our center.

| Patients
We included all consecutive patients with R/R aBCL who received bendamustine LD and were treated with commercial axi-cel at City of Hope (COH) between July and December 2022 during which there was a fludarabine shortage, and all patients who received Flu/Cy LD and commercial axi-cel at COH between January and December 2022 as a reference cohort.Patients' demographics and outcomes were collected through electronic chart review.The study was approved by the COH Institutional Review Board (IRB no.22440).

| Key assessments/measurements
Refractory disease was defined as progressive or stable disease as best response to any line of therapy, or progression within 12 months of transplant, as per ZUMA-1. 10CAR T-cell treatment-emergent toxicities cytokine release syndrome (CRS) and immune effector cell associated neurotoxicity syndrome (ICANS) were graded according to the American Society for Transplantation and Cellular Therapy (ASTCT) consensus criteria. 11All other adverse events were graded by CTCAE v.5.0 criteria.Efficacy was measured by overall response rate (ORR) and complete response (CR) rate.ORR was defined as rate of patients who achieved CR or partial response (PR), as assessed using the Lugano 2014 criteria. 12Overall survival (OS) was defined as the time from CAR T cell infusion to death due to any cause.PFS was defined as the time from CAR T cell infusion to disease progression/relapse or death from any cause, whichever was observed first.

| Statistical considerations
Demographic and disease characteristics were summarized using descriptive statistics including median (range) for continuous variables and count (percentage) for categorical variables.Odds ratios with 95% CI were reported by fitting logistic regression models (one for each of CRS [any grade], CRS [grade ≥3], ICANS [any grade], and ICANS [grade ≥3]) to evaluate the impact of bendamustine relative to Flu/Cy.Adjusted odds ratios with 95%CI were reported by fitting logistic regression models (one for each of prolonged grade ≥3 neutropenia beyond 28 days, grade ≥3 thrombocytopenia, grade ≥3 anemia, infectious complications, and transfusion dependency at 28 days) to evaluate the impact of bendamustine relative to Flu/Cy after adjustment for baseline CAR-HEMATOTOX score (0-1 vs. ≥2).Clopper-Pearson exact method was used to construct the 95% confidence intervals for ORR and CR rate.The Kaplan-Meier method was applied to estimate PFS and OS.Two multivariable Cox proportional hazards models (one for PFS and the other for OS) were fitted to evaluate the impact of bendamustine relative to Flu/Cy LD after adjustment for number of prior lines of therapy (1 vs. ≥2) and refractory disease (no vs. yes).p-values were two-sided with a significance level of 0.05.Statistical analyses were performed in R version 4.2.0 (R Foundation, Vienna, Austria).

| Patient characteristics
We included 27 patients with R/R aBCL who received bendamustine-based LD prior to axi-cel between July and December 2022 when there was a fludarabine shortage.A COH reference cohort of 42 patients with R/R aBCL who received Flu/Cy LD and axi-cel between January and December 2022 were also included.In the bendamustine cohort, 5 (19%) patients received bendamustine alone as LD, while 22 (81%) received bendamustine with rituximab.
Baseline characteristics are summarized in Table 1.In the bendamustine and Flu/Cy-LD cohorts, the median age at infusion were 64 (range: 20-83) years and 62 (range: 35-86) years, 19 (70%) and 37 (88%) received ≥2 prior lines of therapy, and 24 (89%) and 27 (64%) had disease refractory to any line of treatment or relapse within 12 months of transplant.The two LD cohorts were balanced for age, disease stage, presence of bulky disease (>5 cm), International Prognostic Index (IPI) risk score, receipt of bridging, and frequency of elevated pre-LD lactate dehydrogenase (LDH), ferritin, or C-reactive protein.
Relative to patients treated with Flu/Cy, patients treated with bendamustine LD did not show an increase in the hazards  ≥2) (Tables 2A and 2B).

| Hospitalization
Twelve (44%) patients received bendamustine LD and CAR T-cell infusion in the outpatient setting, while 6 (14%) patients received Flu/Cy LD in the outpatient setting.As more patients received LD and axi-cel infusion as outpatient in the bendamustine than Flu/Cy-LD cohort, the median length of hospital stay by day 30 post-infusion in each cohort was 15 days (range 0-30) and 21 days (range: 0-30), respectively.In patients who received bendamustine LD and axi-cel infusion as an outpatient, 10 of 12 (83%) patients required hospital admission with a median length of hospital stay by day 30 of 11.5 days (range: 5-15), for management of CRS (9 ≤ G2, 1 G3).In addition, ICANS occurred in two patients (all ≥G3), and infection occurred in two patients (all ≥G3).The median time between axi-cel infusion and hospital admission was 4 days (range: 0-5), and four (40%) of the patients were admitted on day +3 or earlier.As for the patients receiving Flu/Cy LD and axi-cel infusion outpatient, four out of six (67%) required hospital admission with a median length of hospital stay by day 30 of 10.5 days (range: 2-27), including three for management of CRS (all <G3) and one for G3 infection.Additionally, G3 ICANS and G3 infection occurred in a patient admitted for CRS.The median time between axi-cel infusion and hospital admission was 6.5 days (range: 3-14), with one (25%) patient admitted on day +3 or earlier.

| DISCUSSION
In this study, we reported our center's experience of using bendamustine-based LD in patients with R/R aBCL receiving commercial axi-cel therapy, and compared the efficacy and safety to standard Flu/Cy LD.We did not observe differences in response rates between the two LD regimens cohorts despite a higher proportion of patients with refractory disease in the bendamustine cohort, nor differences in PFS or OS after adjustment for prior lines of therapy and refractory disease at baseline.The observed response rates and 6-month PFS in the bendamustine-LD cohort (ORR 78%, CR 48%, 6-month PFS 44%) were in line with reported outcomes in ZUMA-1 10 (ORR 82%, CR 54%, 6-month PFS 49%) and several real-world studies reporting efficacy of axi-cel use for R/R aBCL in the US (ORR 70%-82%, CR 50%-64%, 6 month PFS 41% 14,15 ), and UK/Europe (ORR 74%-77%, CR 42%-52%, 6-month PFS $45%-49.5% 16,17 ).Bendamustine LD was associated with a more favorable safety profile with lower incidence of ICANS and severe neutropenia compared with Flu/Cy.Hospital utilization tended to be lower among bendamustine-LD recipients, with a higher proportion of patients receiving LD and axi-cel infusion outpatient in comparison with that of Flu/Cy cohort.
The type and intensity of LD prior to CAR T-cell therapy are important factors affecting CAR T-cell expansion and persistence. 2,3Prior studies have shown that suboptimal exposure to fludarabine was associated with poorer relapse-free survival, 18,19 while higher intensity Flu/Cy LD led to favorable cytokine profiles conducive for CAR T-cell expansion in a subset of patients, which was associated with PFS benefit. 2 A recent meeting abstract reported that while Flu/Cy LD induced greater increase in proliferative cytokines (IL-7, IL-2, IL-15) than bendamustine LD, cytokines associated with ICANS (IL-8, monocyte chemoattract protein-1) were also enriched, leading to more toxicity albeit with similar response rates after axi-cel. 20Hence, LD regimens may need to be individualized to balance efficacy and toxicity outcomes in CAR T-cell therapy, but robust biomarkers for predicting efficacy of LD are still under development.Strati et al. reported that change in ALC after LD was prognostic for survival after axi-cel and may potentially be used as a surrogate marker for the biologic effect of LD chemotherapy. 21We observed that both bendamustine and Flu/Cy LD were effective at peripheral blood lymphodepletion and the difference in ALC pre-LD and on day 0 were comparable among patients receiving bendamustine and Flu/Cy LD.Peripheral blood ALC is an easily accessible test, and if prospectively validated as a prognostic marker, has potential to identify patients who may benefit from LD intensity modulation to optimize outcomes after CAR T-cell therapy.
Although our analysis was not designed to directly compare the efficacy of LD regimens, in this unmatched patient population with a median follow-up of 6 months, ORR, PFS, and OS were comparable among patients receiving bendamustine or Flu/Cy LD.While unadjusted survival curves deviate modestly in favor of Flu/Cy LD, no differences in PFS and OS were detected after adjusting for key baseline differences between the two cohorts.We note that median PFS estimates of the bendamustine cohort appear shorter than ZUMA-1 (median 3.5 months at 6-month follow-up vs. 5.8 months at 15-month follow-up 10 ) and several recent real-world studies of axi-cel in R/R aBCL (median PFS 5.5 months to 8.6 months), 14,16,[22][23][24] but this end point may have been influenced by frequency of response assessments and the shorter follow-up of our cohort compared with other studies.
Future analyses, such as incorporation of propensity score matching, along with accumulation of real-world experience, are expected to offer insights and clarity regarding utilization of bendamustine LD for axi-cel.
We observed lower odds of any grade ICANS among recipients of bendamustine LD compared with those treated with Flu/Cy LD.This finding is further supported by lower median dose of corticosteroid used.In contrast, CRS incidence and severity were similar.In realworld evidence (RWE) for patients treated with Flu/Cy LD and axi-cel, Jacobson et al. reported all grade/≥G3 ICANS rate of 55%/24% from CIBMTR data, 22 while European registries reported all grade/≥G3 ICANS rates of 42%-49%/14%-18%. 17,23,24Thus, our bendamustine cohort also exhibited a favorable comparison in the rate of any grade ICANS to recent RWE, supporting the lower risk of neurotoxicity with bendamustine LD compared with Flu/Cy LD.Of note, mitigation strategies for axi-cel-associated toxicity were similar among the two LD cohorts, with prophylactic dexamethasone in high-risk patients, early consideration of steroids for grade 1 ICANS, and use of tocilizumab (with early steroids) for prolonged (>24 h) grade 1 CRS or at onset of grade 2 CRS.Additionally, receipt of bridging, which may decrease CRS and ICANS risk by reducing tumor burden and pretreatment inflammation, 25 was also similar between bendamustine and Flu/Cy LD recipients.With regard to hematologic toxicity, severe neutropenia was observed in a higher proportion of patients after Flu/Cy compared with bendamustine LD, though the myelosuppressive effect of Flu/Cy was not protracted and this did not translate to higher rates of prolonged neutropenia or ≥G3 infectious complications.Overall, our findings were consistent with those from a previous study comparing bendamustine and Flu/Cy LD for tisa-cel, and support a favorable safety profile of bendamustine LD compared with Flu/Cy-LD. 6perience in administering axi-cel outpatient is limited due to high rates of CRS and ICANS associated with CAR T-cell constructs with a CD28 costimulatory domain. 26To our knowledge, our cohort of 18 patients who underwent treatment and monitoring outpatient following axi-cel administration is the largest published to date.We did not observe higher incidence of all-grade and ≥G3 toxicities with outpatient compared with inpatient axi-cel infusion.Notably, 4 of 18 (22%) patients completed 30-day monitoring without admission, and early hospitalization rates within 3 days of axi-cel infusion occurred in 5 of 18 (28%) patients.Moreover, the median length of inpatient stay after axi-cel infusion by day 30 were 20 days (range:12-30) and 8 days (0-27) in patients receiving inpatient and outpatient axi-cel infusion, respectively.Our results suggest that axi-cel infusion in an outpatient setting can be feasible and safe with close patient monitoring and proper protocols for management of CRS, ICANS, and sepsis, which is important given limited hospital resources.
Further inference from our study results is limited by its retrospective nature, which cannot preclude unintended bias in LD selection, and small sample size, which did not allow matching.While we performed adjustment to account for imbalance in patient characteristics when evaluating associations between the LD regimens and safety and survival outcomes, there may be residual confounding.Secondly, heterogeneity in receipt of bendamustine versus bendamustine with rituximab for LD may have influenced efficacy outcomes, although numerically there were no differences in response rates between the two groups.Addition of rituximab to axi-cel in ZUMA-14 elicited high CR rates of 65% and ORR of 88% with no new safety signals, 27 and further studies with a large sample size are needed to evaluate whether addition of rituximab to the LD regimen improves efficacy of CD19 CAR T-cells.Thirdly, patient samples were also not collected for cytokine profiling or measurement of axi-cel expansion in view of its retrospective nature.Lastly, a longer follow-up is warranted to fully evaluate PFS and OS after axi-cel in recipients of bendamustine LD, especially considering long-term follow-up from JULIET reporting inferior survival outcomes of tisa-cel after bendamustine LD in unadjusted analysis. 8 In conclusion, our study evaluating the efficacy and safety profile of bendamustine LD relative to Flu/Cy LD followed by axi-cel in patients with R/R aBCL showed comparable efficacy outcomes.The lower rate of ICANS after bendamustine LD may facilitate administration of axi-cel in the outpatient setting, which shortens overall admission days in comparison with that of axi-cel infusion in the inpatient setting, supporting its cost-effectiveness value. 28The results from our study need prospective confirmation and validation through other robust RWE studies.These results also warrant a prospective evaluation of bendamustine-based LD regimen for SOC CAR T-cell use in patients with lymphoma.
Bendamustine LD comprised of bendamustine (90 mg/m 2 /dose) administered on days À5 and À4, with or without rituximab.Rituximab (375 mg/m 2 /dose) was given once every 28 days starting on day À5 for a total of up to six doses, based on the treating physician's discretion.Standard Flu/Cy LD consisted of a combination of fludarabine (30 mg/m 2 /dose) and cyclophosphamide (500 mg/m 2 /dose) administered on days À5, À4, and À3.The choice of bendamustine rather than Flu/Cyc LD for axi-cel was based on a standardized medical review and rationing of fludarabine supply by Clinical Cellular Immunotherapy Committee at COH. Consideration factors include prior history of bendamustine resistance, availability of fludarabine supply each week, patients' and physicians' preference of using fludarabine, and ability of patient to delay the start of CAR T-cell therapy in patients without rapidly progressing or symptomatic disease.
Abbreviations: CRS, cytokine release syndrome; ICANS, immune effector cell-associated neurotoxicity syndrome.a Percentage% (Count); Median (Range).b The odds of experiencing CRS or ICANS among the Benda cohort relative to the odds among the Flu/Cy cohort by fitting a logistic regression model for each toxicity outcome with 95% confidence intervals constructed using robust standard errors.c The odds of experiencing the hematological toxicity among the Benda cohort relative to the odds among the Flu/Cy cohort after adjustment for baseline CAR-Hematotox score (0-1 vs. ≥2) by fitting a logistic regression model for each hematological toxicity outcome with 95% confidence intervals constructed using robust standard errors.d Patient died at day 1 after infusion.e ANC nadir not measured.fOdds ratio cannot be calculated due to no variance in the Flu/Cy cohort.g In the Flu/Cy cohort, two patients died prior to day 28 after infusion (at day 1 and day 24, respectively).h In the Benda cohort, one patient died prior to day 28 after infusion (at day 23), and another two patients did not have ANC measured.
Nevertheless, our early results showing comparable efficacy and safety of bendamustine LD relative to Flu/Cy LD supports further exploration of bendamustine-based LD for CD28-based CAR T-cell products.

1
Baseline characteristics of patients with aggressive B-cell lymphoma receiving bendamustine versus fludarabine/cyclophosphamide lymphodepletion.