Benefit of axicabtagene ciloleucel versus chemoimmunotherapy in older patients and/or patients with poor ECOG performance status with relapsed or refractory large B‐cell lymphoma after 2 or more lines of prior therapy

Axicabtagene ciloleucel (axi‐cel) in trials has demonstrated favorable efficacy compared with historical controls after ≥2 lines of therapy for the treatment of relapsed or refractory (R/R) large B cell lymphoma (LBCL). Herein, we compared the real‐world effectiveness of axi‐cel with efficacy and effectiveness of chemoimmunotherapy (CIT) in patients aged ≥65 years and patients with Eastern Cooperative Oncology Group performance status (ECOG PS) of 2. A total of 1146 patients treated with commercial axi‐cel for R/R LBCL with ≥2 lines of prior therapy were included from the Center for International Blood and Marrow Transplantation Research prospective observational study, and 469 patients treated with CIT for R/R LBCL after ≥2 lines of prior therapy were included from SCHOLAR‐1 (an international, multicohort, retrospective study). After propensity score matching, at a median follow‐up of 24 months for patients receiving axi‐cel and 60 months for patients receiving CIT, 12‐month overall survival rates were 62% and 28%, respectively (hazard ratio, 0.30 [95% CI, 0.24–0.37]). Objective response rate (ORR) was 76% (complete response [CR] rate 58%) in patients receiving axi‐cel versus 28% (CR rate 16%) for those receiving CIT. A 57% difference in ORR (55% difference in CR rate) favoring axi‐cel over CIT was observed among patients aged ≥65 years. Increased magnitude of benefit in response rates for axi‐cel versus CIT was also observed among patients with ECOG PS = 2. These findings further support the broader use of axi‐cel in older patients and patients with ECOG PS = 2 with R/R LBCL.


| INTRODUCTION
5][6][7] Additionally, comorbidities associated with advanced age and poor Eastern Cooperative Oncology Group performance status (ECOG PS) are common exclusion criteria in clinical trials, making the efficacy of treatment options in these populations, especially in the real-world setting, unclear. 8For these reasons, it is crucial to understand the effectiveness of available treatment options for R/R LBCL among these patient groups.
Axicabtagene ciloleucel (axi-cel) is an autologous anti-CD19 chimeric antigen receptor (CAR) T-cell immunotherapy with a CD28 costimulatory domain that provides rapid and strong expansion of T cells, and reprograms T cells to trigger target-specific cytotoxicity of cancer cells. 90][11] In a subgroup analysis of the Phase 1/2 ZUMA-1 study, which examined axicel in refractory LBCL after 2 or more lines of prior therapy by age, axi-cel induced durable responses with a manageable safety profile in patients aged ≥65 years, with a median event-free survival of 12.5 months after >5 years of follow-up. 3,12Additionally, in a subgroup analysis of ZUMA-7, a Phase 3 study of axi-cel versus standardof-care in second-line (2 L) LBCL, axi-cel demonstrated superior overall survival (OS) over standard-of-care in patients aged ≥65 years (hazard ratio [HR], 0.517 [95% CI, 0.277-0.964]),along with improved patient-reported outcomes in patients aged ≥65 years. 13recent analysis based on real-world data from the observational database of the Center for International Blood and Marrow Transplant Research (CIBMTR) was performed on patients receiving axi-cel in the third-or later-line settings.This analysis by Jacobson et al. showed that treatment with commercial axi-cel demonstrated outcomes consistent with results from the ZUMA-1 study. 14Additionally, axi-cel showed a favorable objective response rate (ORR) among patients aged ≥65 years compared with patients aged <65 years after multivariable adjustment of other prognostic factors. 14In terms of survival outcomes, SCHOLAR-1, an international, multicohort, retrospective study of patients with R/R LBCL treated with CIT, showed similarly poor 24-month OS rates in patients aged <65 years and those aged ≥65 years (24-month OS rate, 20% [95% CI, 16-23] and 19% [95% CI, 11-29], respectively) and significantly lower median OS in patients with ECOG PS ≥2 versus patients with ECOG PS <2 (HR, 2.1; p < .0001). 15Here, we examine the real-world effectiveness of axi-cel compared with CIT in patients with R/R LBCL from data collected for a post-authorization safety study using the CIBMTR and SCHOLAR-1 datasets after ≥2 lines of therapy, and further explore the magnitude of the effectiveness benefit seen with axi-cel among older patients and patients with an unfavorable ECOG PS.

| Patients and study design
The CIBMTR is a collaborative working group of more than 500 treatment centers worldwide managed by the Medical College of Wisconsin and the National Marrow Donor Program (NMDP).Detailed information on patient, treatment and disease characteristics, and demographics were longitudinally reported by participating centers.
Integrity and quality of data were monitored at different levels, including onsite audits, automated and manual checks for discrepancies.
Patients signed informed consent to share data with the CIBMTR for research studies, and utilization of these data for research was overseen by the NMDP central institutional review board.Real-world data from CIBMTR was used for the development of a post-authorization safety study (PASS) to prospectively capture safety and effectiveness information for axi-cel recipients with R/R LBCL after 2 or more lines of prior therapy.A total of 1500 patients receiving commercial axi-cel for R/R LBCL between October 2017 and August 2020 at 79 centers were enrolled in the PASS.The SCHOLAR-1 analysis collected patient-level data from patients treated with CIT for R/R LBCL after 2 or more lines of prior therapy from 4 databases with enrollment years from 2001 to 2014. 15Full inclusion and exclusion criteria for SCHOLAR-1 were previously reported. 15igible patients for this analysis received commercial axi-cel or CIT as the third or later line of therapy for LBCL.Patients from SCHOLAR-1 who were indexed after 1 prior line of therapy were included and reindexed after ≥2 lines of therapy if they had a third line of treatment and were refractory or relapsed after the second line of therapy, or received autologous stem cell transplantation (ASCT) just before index line.Any patients originally indexed after 1 prior line of therapy who did not meet these criteria were excluded.Exclusion criteria for this study included axi-cel received in noncommercial settings and prior history of allogeneic stem cell transplant.

| Treatment and study endpoints
In the study population treated with axi-cel derived from the CIBMTR database, patients received post-approval axi-cel per institutional practice as part of the prospective observation study.Participating sites were responsible for completing a data collection form at predetermined time points.Treatment in SCHOLAR-1 was previously described. 15fectiveness endpoints assessed in this analysis included ORR, CR rate, and OS.Response was assessed using Lugano 2014 response criteria for patients who received axi-cel while response assessments varied for patients in SCHOLAR-1. 16Safety and/or tolerability were not examined in this analysis because safety data were not collected for SCHOLAR-1.

| Statistical analysis
All patients who received axi-cel or CIT who met eligibility criteria (full analysis set) were assessed for baseline characteristics.Those with available response data were included in the response rate analysis set and those with available OS data were included in the survival analysis set.In this analysis, propensity score matching (PSM) was performed to reduce imbalanced distribution of baseline characteristics between the populations treated with axi-cel and CIT.PSM was performed separately for the response rate and survival analysis sets due to the substantial difference in sample sizes of the 2 analysis sets, as 1 of the 4 data sources for SCHOLAR-1 collected response data only up to the second line of salvage therapy and thus had to be excluded altogether from the response rate analysis set.PSM was performed with a 2:1 ratio and nearest neighbor matching, with a caliper (maximum permitted difference in propensity score between matched patients) of 0.2.Variables considered in the logistic regression model to obtain the propensity score included age (<65 vs. ≥65 years), sex (female vs. male), ECOG PS (0-1 vs. ≥2 vs. unknown), disease histology (diffuse LBCL vs. primary mediastinal B-cell lymphoma vs. transformed follicular lymphoma), disease stage at initial diagnosis (I to II vs. III to IV vs. unknown), refractory to all prior lines of therapy (yes vs. no), and prior ASCT (for the response rate analysis set only; yes vs. no).A stepwise variable selection process was used for model building, using a p value cutoff of 0.2 for variables to enter the model and 0.05 for variables to stay in the model.The significant variables in the model were then subsequently used for PSM for the response rate and survival analysis sets.
After PSM, proportions and 95% Clopper-Pearson CIs were calculated for ORR and CR rate, and Kaplan-Meier estimates were used for OS.Odds ratios (ORs) for ORR and CR and HR for OS were estimated by multivariate logistic regression and Cox regressions, respectively, for axi-cel versus CIT to further adjust for residual confounding.
All analyses were carried out for the full PSM population and were repeated for the subgroups of the full PSM population aged <65 and ≥65 years at time of treatment and with ECOG PS <2 and ECOG PS = 2 at time of treatment.For the subgroup analyses by ECOG PS, patients with ECOG PS >2 were excluded as there were no patients with ECOG PS >2 in the axi-cel arm and only 3 patients in the CIT arm.
Conventional multivariable regression models based on the full, unmatched population were also performed as sensitivity analyses, and direct adjusted survival functions were generated. 17,18All analyses were performed in SAS 9.4 (M1).

| Systematic literature search of contemporary standard-of-care therapies
In order to account for contemporary non-CAR T-cell therapies that were not used widely until after the completion of the SCHOLAR-1 study, a systematic literature review (SLR) was performed that identified clinical and real-world studies that examined the efficacy or effectiveness of these therapies.The contemporary treatments included in this review are epcoritamab, glofitamab, loncastuximab tesirine, and selinexor in the third line or later setting along with polatuzumab vedotin in combination with bendamustine and rituximab (pola-BR) and tafasitamab plus lenalidomide in the 2 L or higher setting.A PRISMA-compliant SLR was developed to identify relevant studies.MEDLINE ® , Embase and the Cochrane Library were searched on 13 January 2023 for relevant studies published from 2017 to the search date.All major congresses from 2020 onwards were captured via Embase except for the American Society of Hematology (ASH) 2022 congress, for which the conference proceedings were searched by hand.Congress abstracts and abstracts from full publications were screened against agreed inclusion and exclusion criteria (Table S1).The full texts of the selected publications were then reviewed to confirm their eligibility.Both screening stages were performed in a single-blind manner and any uncertainties on screening decisions were discussed with a second reviewer.Data were extracted into a data extraction table in Excel, and each extracted data point was checked by a second person.
Quality assessment, or risk of bias assessment, was performed according to study type.

| Patients
A total of 1146 patients from 79 US centers treated with commercial axi-cel for R/R LBCL were included for this analysis (Figure S1).Additionally, 469 patients treated with CIT for R/R LBCL were included from the SCHOLAR-1 dataset.The median follow-up for patients receiving axi-cel was 24.5 months (95% CI, 24.3-24.9),and the median follow-up for patients receiving CIT was 59.8 months (95% CI, 52.3-65.0)based on the reverse Kaplan-Meier method. 19 After PSM, 493 patients who received axi-cel and 289 patients who received CIT were included in the response rate analysis set, while 659 patients who received axi-cel and 406 patients who received CIT were included in the survival analysis set.Subgroup analysis by age (<65 years vs. ≥65 years) was also performed in these 2 analysis sets.In the subgroup analysis by ECOG PS (<2 vs. 2), after PSM, 403 patients who received axi-cel and 207 patients who received CIT were included in the response rate analysis set, while 559 patients who received axi-cel and 323 patients who received CIT were included in the survival analysis set due to further exclusion of patients with missing ECOG PS.Baseline characteristics for patients in the PSM population were largely balanced (standard mean deviation [SMD] range for response analysis set, À0.19-0.06;SMD range for survival analysis set, À0.09-0.06)except for number of lines of prior therapy, of which a markedly greater proportion of axi-cel recipients had ≥3 prior lines compared with CIT recipients (67% vs. 5% in the response rate analysis set and 68% vs. 4% in the survival analysis set; Table 2).

| Sensitivity analysis
Sensitivity analyses performed using conventional multivariable logistic and Cox regressions in the full, unmatched patient population demonstrated results consistent with the PSM analysis (Figure S2).Among all patients, those who received axi-cel had a significantly higher ORR extraction and reporting (Figure S3).Twenty additional abstracts from ASH 2022 were identified by hand search, resulting in 104 publications included in the results.Real-world studies were identified for pola-BR and tafasitamab plus lenalidomide, but only clinical trial publications were available for all other treatments included in the search.
Detailed results are reported in the Supplemental Appendix.Briefly, ORRs with treatments from clinical trials in the 3 L or later setting ranged from 28% (selinexor) to 63% (epcoritamab and glofitamab), and CR rate ranged from 12% (selinexor) to 52% (glofitamab; Table S5).Medians of OS with these therapies ranged from 9.1 months (selinexor) to 11.5 months (glofitamab) while median OS was not reached in one study (epcoritamab; Table S5).Real-world studies in the 3 L or later setting were only identified for pola-BR, whereby the ORR ranged from 21% to 22%, and the CR rate was 7%.Generally, clinical outcomes were numerically lower in real-world studies versus interventional studies.
A subgroup analysis of studies that examined outcomes by age in the 3 L or later setting found that ORRs in older patients ranged from 24% (selinexor; aged ≥65 years) to 52% (loncastuximab tesirine; aged ≥75 years; Table S6).The ORR in younger patients in this setting ranged from 31% (selinexor; aged <70 years) to 49% (loncastuximab tesirine; aged <65 years).Median OS was 7.8 months in older patients (aged ≥65 years or ≥70 years) treated with selinexor.Median OS data in older patients were unavailable for the other therapies examined in this analysis.
In an additional subgroup analysis by ECOG PS in the 3 L or later setting, it was found that patients with ECOG PS ≥1 treated with glofitamab had an ORR of 61% and a CR rate of 55%.Patients with ECOG PS = 0 had an ORR of 65% and a CR rate of 50%.

| DISCUSSION
The treatment landscape for older and more frail patients with R/R LBCL can be more limited than that for the general population due to to evaluate their effectiveness versus standard-of-care in this population.It is also key to evaluate the benefit of these therapies relative to standard-of-care using real-world data, as the real-world treatment population will include patients who were ineligible for the therapy in a trial setting (eg., patients with ECOG PS = 2 were excluded from the ZUMA-1 and ZUMA-7 trials). 14In this analysis, axi-cel CAR T-cell therapy demonstrated superior response and survival versus CIT among older patients and those with unfavorable ECOG PS with R/R LBCL in a real-world treatment setting.
This real-world analysis showed that axi-cel provides superior benefit versus CIT to patients with R/R LBCL regardless of age.ORR and CR rates in this analysis were comparable among patients aged <65 and ≥65 years, consistent with prior real-world and clinical analyses of axi-cel and CIT by age. 3,14,15,20Consistent responses across age groups were also observed in clinical studies of patients with R/R LBCL treated with tisagenlecleucel and lisocabtagene maraleucel. 21,22e current analysis further demonstrated favorable response rates for axi-cel treatment relative to CIT among patients of all ages with R/R LBCL in a real-world setting.4][25] Notably, a larger relative benefit in ORR (57% difference for patients aged ≥65 years; 46% difference for patients aged <65 years) and CR (55% difference for patients aged ≥65 years; 39% difference for patients aged <65 years) over CIT was observed among patients aged ≥65 years who received axi-cel than among patients aged <65 years, suggesting that the clinical benefit of axi-cel versus CIT is more pronounced among older patients.Additionally, patients with ECOG PS = 2 who received axicel showed a larger relative benefit in ORR over CIT versus patients with ECOG PS <2 (52% difference for patients with ECOG PS = 2; 41% difference for patients with ECOG PS <2).ECOG PS ≥2 has been previously associated with inferior efficacy outcomes among patients treated with axi-cel versus patients with ECOG PS <2. 14,26This analysis suggests that axi-cel is a more effective treatment option than CIT for this difficult-to-treat population.
The 12-month OS rate among all patients with R/R LBCL who received axi-cel in this analysis was comparable to prior clinical and real-world studies 3,27 and similar to prior studies of lisocabtagene maraleucel while being higher than prior studies of tisagenlecleucel. 21,22OS was longer in patients who received axi-cel versus patients who received CIT, as was previously demonstrated in the analyses comparing ZUMA-1 and SCHOLAR-1. 23,24Specifically, OS was significantly prolonged in patients receiving axi-cel versus patients receiving CIT in all the subgroups analyzed except among patients with ECOG PS = 2, where OS was only numerically prolonged.However, it should be noted that this subgroup of patients were not available to be examined in this analysis, but prior studies have shown that axi-cel safety outcomes for older patients and patients with unfavorable ECOG PS are manageable. 28,29 conclusion, these results demonstrated the clinical benefit of

F I G U R E 1
Survival Curves for Adjusted OS for All Patients, the Subgroups of Patients Aged <65 Years and ≥65 Years, Patients With ECOG PS <2 and ECOG PS = 2, and Hazard Ratios.Panel A shows the adjusted OS estimates and HR in all patients.Panel B shows the adjusted OS estimates and HR and patients aged <65 and ≥65 years.Panel C shows the adjusted OS estimates and HR in patients with ECOG PS <2 and ECOG PS = 2. Panel D shows HR for all patients and all subgroups.axi-cel, axicabtagene ciloleucel; CIT, chemoimmunotherapy; ECOG PS, Eastern Cooperative Oncology Group performance status; HR, hazard ratio; KM, Kaplan-Meier; NE, not estimable; No., number; OS, overall survival.[Color figure can be viewed at wileyonlinelibrary.com] increased prevalence of comorbidities.As such, when new treatment options including CAR T-cell therapy become available, it is essential

F I G U R E 2
Adjusted Odds Ratios for ORR and CR Rate for All Patients, and the Subgroups of Patients Aged <65 Years and ≥65 Years and Patients With ECOG PS <2 and ECOG PS = 2. Panel A shows the ORR and OR for all patients, as well as the subgroups of patients aged <65 and ≥65 years and patients with ECOG PS <2 and ECOG PS = 2. Panel B shows the CR rates and OR for all patients, as well as the subgroups of patients aged <65 and ≥65 years and patients with ECOG PS <2 and ECOG PS = 2. axi-cel, axicabtagene ciloleucel; CIT, chemoimmunotherapy; CR, complete response; ECOG PS, Eastern Cooperative Oncology Group performance status; OR, odds ratio; ORR, objective response rate.[Color figure can be viewed at wileyonlinelibrary.com]had a limited patient pool (n = 33 for axi-cel; n = 27 for CIT).This suggests that patients with R/R LBCL who receive axi-cel can achieve improved survival versus those who receive CIT, even among older patients who often have significant comorbidities.An SLR of contemporary non-CAR T-cell therapies for R/R LBCL suggested that the relative benefit of axi-cel versus standard-of-care therapy is retained even when including newer treatment options.The contemporary therapies examined in this study included an anti-CD19 monoclonal antibody-containing therapy (tafasitamab + lenalidomide), bispecific antibodies (epcoritamab and glofitamab), antibody-drug conjugate-containing therapies (pola-BR and loncastuximab tesirine), and a selective inhibitor of nuclear export (selinexor).While several of these therapies showed improvements in efficacy versus those examined in SCHOLAR-1, response rates still lagged numerically behind those seen in patients receiving axi-cel, particularly in the real world, and/or 3 L or later settings.Additionally, median OS for patients receiving contemporary therapies appeared shorter than that of patients receiving axi-cel in all real-world studies examined.In a subgroup analysis by age and ECOG PS, the relative effectiveness and survival benefit of axi-cel versus standard-of-care therapy in older patients was retained versus these contemporary therapies.As in all non-randomized studies, limitations of the SCHOLAR-1 and PASS studies include variability in patients' treatment periods and response assessments, and differences in clinical practice and supportive care across the studies and across the time periods included in the analysis.It should also be noted that comorbidity data were not available in the SCHOLAR-1 dataset, precluding the inclusion of this covariate in PSM.Other potential biases in patient selection could not be fully controlled, including any patient characteristics that may lead to a preference of either CAR T-cell therapy or CIT, especially among older patients.For example, a greater proportion of patients in the axi-cel arm had ≥3 lines of prior therapy versus the CIT arm, which may have favored the CIT arm.Owing to the historical timeframe of the data collected for the SCHOLAR-1 study, some contemporary therapies were not examined as comparators for patients receiving axi-cel.However, the SLR of contemporary therapies for R/R LBCL suggested that while patient outcomes for non-CAR T-cell therapies have improved since SCHOLAR-1, they are still numerically inferior to those observed for patients receiving axi-cel.Further, the SCHOLAR-1 dataset had no race/ethnicity information and thus we cannot know if there is a patient imbalance between axi-cel and CIT based on these characteristics.Geographic distribution of patients was also potentially imbalanced between treatment arms, as SCHOLAR-1 includes international patients while the PASS included only US patients.Additionally, analysis of OS was potentially limited by the relatively short follow-up period available in this analysis and the substantial confounding factor of death from other causes in patients aged ≥65 years and patients with ECOG PS = 2.This analysis of patient aged ≥65 years includes a relatively small number of patients and should be interpreted with caution.Additionally, patients aged ≥65 years who received axi-cel demonstrated higher response than patients aged <65 years, which may suggest some degree of patient selection for CAR T-cell therapy by the treating physicians among the older patient population.It also would have been ideal to examine the subgroup of patients aged ≥70 years in each treatment arm, but this was precluded by the small number of patients aged ≥70 years treated with CIT in SCHOLAR-1.However, patients aged ≥70 years treated with axi-cel had comparable effectiveness outcomes to patients aged ≥65 years, suggesting that axi-cel is similarly effective in this patient population.Finally, safety data treatment with axi-cel over CIT in patients with R/R LBCL in the realworld setting.In patients aged ≥65 years, the magnitude of response and length of survival observed with axi-cel versus CIT was numerically greater than that of the overall population, suggesting that this population especially may benefit from treatment with axi-cel.A clinical benefit for patients treated with axi-cel versus CIT was also observed in patients with ECOG PS = 2, a similarly difficult-to-treat patient population.Taken together, these data support axi-cel CAR T-cell therapy as an effective treatment option for patients with R/R LBCL who are older and/or those with unfavorable ECOG PS.AUTHOR CONTRIBUTIONS Participated in study design: MAL, HW, ZH, HM, MCP, SAS, CS, HX.Collected data: HW, ZH, SSN, JW, MCP, HX.Completed the statistical analysis: HW, ZH, CD, PF.Provided logistical management of the study: DM.All authors were involved in the interpretation of the data and writing of the article, and provided final approval to submit it for publication.

Table 1 )
than patients who received CIT.
Baseline characteristics for the response rate and survival analysis populations after PSM.Note: The variables selected for PSM for the Response Rate Analysis set included age, ECOG PS, disease histology, disease stage, refractory to all prior lines of therapy, and prior ASCT.The variables used for PSM for the Survival Analysis set included age, ECOG PS, disease histology, disease stage, and refractory to all prior lines of therapy.Abbreviations: ASCT, autologous stem cell transplantation; axi-cel, axicabtagene ciloleucel; CIT, chemoimmunotherapy; DLBCL, diffuse large B-cell lymphoma; ECOG PS, Eastern Cooperative Oncology Group performance status; N/A, not applicable; PD, progressive disease; PSM, propensity score matching; SD, stable disease; SMD, standard mean deviation.