Three-year follow-up analysis of first-line axicabtagene ciloleucel for high-risk large B-cell lymphoma: the ZUMA-12 study Free

Standard first-line treatment of large B-cell lymphoma (LBCL) with anti–CD20-based chemoimmunotherapy results in a high rate of relapse in patients with high-risk characteristics, including those with a high International Prognostic Index (IPI) score (≥3) and/or a diagnosis with high-grade B-cell lymphoma (with gene rearrangements in MYC and BCL2 and/or BCL6, referred to as double- or triple-hit lymphomas).¹ Treatment strategies to improve outcomes with first-line therapy in high-risk patients have been limited in part due to underrepresentation of patients with high-risk LBCL in clinical trials.² Although positive interim positron emission tomography after 2 cycles of first-line chemotherapy (interim PET2+) is associated with poor survival outcomes, attempts to improve outcomes with chemotherapy intensification have largely been unsuccessful.³ 

Autologous anti–CD19-directed chimeric antigen receptor (CAR) T-cell therapies, such as axicabtagene ciloleucel (axi-cel), have emerged as viable options in LBCL, with approvals in several countries in the second line for patients with early relapse or refractory disease, as well as for those with ≥2 relapses.^4-7 Axi-cel has shown superiority over second-line standard of care with curative intent in patients intended for high-dose chemotherapy with autologous stem cell transplantation, demonstrating significant improvement in overall survival (OS) over platinum-based chemotherapy and autologous stem cell transplantation.⁸ 

ZUMA-12 is a prospective, phase 2, multicenter, single-arm trial evaluating axi-cel as part of first-line therapy in high-risk LBCL, defined by baseline features including double- or triple-hit lymphomas and/or IPI scores ≥3, as well as poor risk functional features determined by interim PET2+. The primary analysis of ZUMA-12 demonstrated a high complete response (CR) rate of 78%, with no safety concerns emerging in this line, supporting axi-cel as a highly effective option as part of first-line therapy for patients with high-risk LBCL.⁹ Here, we report efficacy, durability, and safety outcomes with axi-cel as part of first-line therapy for high-risk LBCL from ZUMA-12 after a median of ≥3 years of follow-up.

ZUMA-12 is a multicenter, single-arm, prospective, phase 2 trial at 7 medical centers across the United States, Australia, and France and is registered at Clinicaltrials.gov/NCT03761056. The full details of the trial centers have been reported previously.⁹ All patients enrolled in the study provided written informed consent, and the study was conducted in accordance with applicable International Conference on Harmonisation Good Clinical Practice Guidelines, the principles of the Declaration of Helsinki, and any applicable local laws and regulations.

Full patient eligibility criteria have been previously reported.⁹ Briefly, eligible patients were aged ≥18 years with histologically confirmed LBCL, including diffuse LBCL (DLBCL)–not otherwise specified, high-grade B-cell lymphoma with MYC and BCL2 and/or BCL6 (double- or triple-hit lymphomas), or high-grade B-cell lymphoma–not otherwise specified. Patients also must have had high-risk disease, defined as double- or triple-hit lymphomas and/or an IPI score of ≥3 at initial time of diagnosis or between initial diagnosis and enrollment. Patients must have had an interim PET2+ with a Deauville PET score of 4 or 5 per Lugano classification. The initial 2 cycles of chemoimmunotherapy must have consisted of an anti-CD20 monoclonal antibody and an anthracycline-containing regimen.¹⁰ Patients must have had an Eastern Cooperative Oncology Group performance status of 0 or 1 at the time of enrollment. There must have been no evidence of central nervous system (CNS) involvement of the lymphoma.

Enrolled patients underwent leukapheresis, followed by lymphodepleting chemotherapy, consisting of fludarabine at 30 mg/m² per day and cyclophosphamide at 500 mg/m² per day on days −5 to −3. Infusion of axi-cel at a target dose of 2 × 10⁶ anti-CD19 CAR T cells/kg occurred on day 0 (flat dose of 2 × 10⁸ cells in patients weighing ≥100 kg), followed by a minimum of 7 days of hospitalization. Optional bridging therapy was administered per investigator discretion before lymphodepleting therapy and could include corticosteroids or high-dose methylprednisolone with rituximab.

Disease response assessments occurred every 3 months up to month 24, and further assessments occurred every 6 months until month 60, then annually thereafter or until disease progression as per standard of care.⁹ Data on targeted adverse events and serious adverse events, including neurologic events, hematologic events, infections, autoimmune disorders, and secondary malignancies, were collected up to 15 years after infusion or until disease progression. Serious adverse events related to axi-cel were reported regardless of time frame. Adverse events were coded via the Medical Dictionary for Regulatory Activities version 23.1 and graded per the National Cancer Institute Common Terminology Criteria for Adverse Events version 5.0. Cytokine release syndrome cases were graded according to a modified version of the system described by Lee et al.¹¹ Biomarker assessments were performed as previously described.⁹ 

The primary end point was investigator-assessed CR. Secondary end points included objective response rate, duration of response (DOR), event-free survival (EFS; time from infusion to earliest date of disease progression, commencement of subsequent new antilymphoma therapy, including stem cell transplantation, or death from any cause), progression-free survival (PFS), OS, incidence of adverse events, relapse with CNS disease, as well as pharmacokinetic assessments. In all enrolled patients, survival estimates were calculated from the time of enrollment.

The final statistical analysis occurred after the last patient completed the 36-month visit. An additional analysis of survival was conducted ≈6 months thereafter. No formal statistical hypotheses were tested in this analysis, and assessments of response and safety were descriptive. Patients evaluable for response consisted of those with centrally confirmed double- or triple-hit lymphomas and/or IPI score ≥3 who received a target dose. Two-sided 95% confidence intervals (CIs) for response were generated using the Clopper-Pearson method. All patients receiving any dose of axi-cel were included in the safety analysis. Time-to-event measurements were calculated using the Kaplan-Meier method.

As reported previously, 42 patients enrolled and underwent leukapheresis between 6 February 2019, and 22 October 2020, with 40 patients receiving axi-cel (see prior report for Consolidated Standards of Reporting Trials [CONSORT] diagram).⁹ As of the data cutoff date of 19 December 2023, treated patients had a median follow-up of 48.5 months (range, 37.1-57.8 months). Patient baseline characteristics were previously reported.⁹ 

Among patients who were evaluable for response (n = 37; confirmed disease type and/or IPI score ≥3 who received axi-cel at the target dose), the median follow-up was 47.0 months (range, 37.1-57.8 months). The CR rate was 86% (95% CI, 71%-95%; n = 32), numerically higher than the primary analysis (Figure 1).⁹ Two patients achieved CR after the data cutoff of the primary analysis, and a third patient was found to be in CR during prior visits after data review. In total, 8 patients (22%) with an initial partial response converted to a CR, and 1 patient (3%) with stable disease converted to a CR by data cutoff. By 6 months after infusion, 70% (95% CI, 53%-84%) of patients achieved a CR (supplemental Figure 1 [available on the Blood website]). All 4 patients with double- or triple-hit lymphomas and an IPI score ≥3 had CR, as previously reported.⁹ Among 6 patients with double- or triple-hit lymphoma only (IPI ≤2), 5 (83%) had a CR. Of 27 patients with IPI score ≥3 only, 23 (85%) achieved a CR. The objective response rate was 92% in response-evaluable patients (n = 34), and 93% in all treated patients (37/40; 88% CR rate; supplemental Table 1).

At data cutoff, 27 of the 37 evaluable patients (73%) were in ongoing response, all of whom were in ongoing CR. When assessed only among those 32 patients with a CR, 84% had an ongoing response. Median DOR was not reached (95% CI, not estimable [NE]-NE), and the 36-month rate of ongoing response was 81.8% (95%, 63.9%-91.4%; Figure 1; supplemental Table 2). Median DOR was not reached among patients who had a best response of a CR, whereas the 2 patients with a partial response as best response had a median DOR of 2 months (95% CI, NE-NE). The 36-month ongoing response rate was 84.4% (95% CI, 66.5%-93.2%) among those who achieved a CR. No patients experienced relapse at 18 months after infusion.

Median EFS was not reached (95% CI, NE-NE; Figure 2). The 36-month estimated EFS rate was 73.0% (95% CI, 55.6%-84.4%), and was consistent among high-risk baseline characteristics (Figure 3). Similarly, medians of PFS and OS were not yet reached, and 36-month estimates were 75.1% (95% CI, 57.5%-86.2%) and 81.1% (95% CI, 64.4%-90.5%; Figure 2), respectively. No relapse with CNS disease was reported at any time on study. Patients who achieved a CR as best response had a 36-month OS estimate of 90.6% (95% CI, 73.7%-96.9%).

Of all 40 treated patients, 32 (80%) were still alive at data cutoff. Among those, 2 had disease progression after a CR, both within 12 months after infusion. One had progressed on subsequent rituximab + daratumumab, followed by a CR to lenalidomide + obinutuzumab. The other had a CR to subsequent radiotherapy, followed by disease progression. One patient was retreated with axi-cel, achieved a CR to retreatment, and received subsequent polatuzumab + bendamustine + rituximab with a best response of progressive disease and subsequent death reported thereafter.

Efficacy outcomes among treated and enrolled patients remained similar to those in response-evaluable patients (supplemental Table 1). The objective response rate in treated patients was 93% (n = 37), with a CR rate of 88%. Estimated duration of response at 36 months was 80.6% (95% CI, 63.5%-90.2%). Estimates at 36 months of EFS, PFS, and OS in treated patients were 72.5% (95% CI, 55.9%-83.7%), 74.4% (95% CI, 57.7%-85.3%), and 80.0% (95% CI, 64.0%-89.5%), respectively. In enrolled patients, 88% achieved a response, with a CR rate of 83%. The 36-month rates of EFS and PFS in enrolled patients were consistent with those in treated patients. The 36-month estimated OS in enrolled patients was 80.1% (95% CI, 64.1%-89.5%). No medians of time-to-event outcomes were reached in treated or enrolled patients.

In a longer-term assessment of survival in response-evaluable patients, after a median of 50.9 months of follow-up (data cutoff, 15 April 2024), median OS remained not yet reached (95% CI, 54.5-NE; supplemental Figure 2). The 48-month estimated OS was the same as that at 36 months (81.1%). One additional patient died after the data cutoff of the 3-year analysis due to pneumonia (54.4 months after infusion).

No new safety signals were observed after the primary analysis.⁹ Grade ≥3 adverse events that occurred after the primary analysis were 1 grade 3 COVID-19 (745 days after infusion), 1 grade 3 COVID-19 pneumonia (458 days after infusion), 1 device-related infection (458 days after infusion), and 2 new malignancies (373 and 744 days after infusion; Table 1). In total, new infections occurred in 2 patients (1 patient had grade 2 and grade 3 COVID-19, and 1 patient had grade 3 COVID-19 pneumonia and grade 3 device-related infection), all serious events not related to axi-cel. No new cytokine release syndrome cases, neurologic events, cytopenias, or hypogammaglobulinemia occurred. In total, 4 patients received immunoglobulin replacement therapy. The previously reported ongoing neurologic event had resolved by data cutoff. ⁹ No tumor lysis syndrome or replication-competent retrovirus was reported any time while on the study. In total, 4 patients had new malignancies: 1 squamous cell carcinoma (477 days after infusion), 1 esophageal adenocarcinoma (373 days after infusion), 1 gastrointestinal stromal tumor (224 days after infusion, event occurred prior to the data cutoff of the primary analysis but was recorded afterward), and 1 leukemia (744 days after infusion; considered related to lymphodepleting chemotherapy; clonal hematopoiesis of indeterminate potential [CHIP] mutation present before CAR T-cell therapy); none was related to axi-cel.

A total of 8 patients (20%) treated with axi-cel died at any time on study. All deaths occurred after at least 3 months postinfusion. Disease progression was the cause of death in 5 patients. Deaths due to nonrelapse mortality (without prior disease progression) were rare, occurring in 2 patients (5%): 1 due to COVID-19 on day 350, as previously reported, and 1 occurring after the data cutoff of the primary analysis due to esophageal adenocarcinoma on day 535. Neither was related to axi-cel.⁹ The final death was due to sepsis, as previously reported.⁹ 

Median peak CAR T-cell levels were 35.7 cells/μL (range, 6.8-560.3 cells/μL) among 27 response-evaluable patients with ongoing response at data cutoff, 45.1 cells/μL (range, 16.0-141.8 cells/μL) among 6 patients who relapsed, and 34.7 cells/μL (range, 34.5-77.9 cells/μL) among 3 nonresponders (Figure 4). Median areas under the curve of CAR T-cell expansion between days 0 and 28 were 368.0 cells/μL × days (range, 74.5-4288.0 cells/μL × days) in those with ongoing response, 413.4 cells/μL × days (range, 153.9-1974.2 cells/μL × days) in those who relapsed, and 566.8 cells/μL × days (range, 433.4-718.8 cells/μL × days) in those with no response. These differences were not statistically significant. By month 24 after infusion, low levels of CAR gene-marked cells were detectable in 41% of patients in ongoing response. B cells were detected by month 12 after infusion among 76.5% of patients in ongoing response and in 50% of patients who had relapsed. Levels of CAR gene-marked cells and B cells were inversely correlated at each time point after infusion (supplemental Figure 3).

Patients with high-risk LBCL remain at a high unmet need, with modest outcomes with standard first-line chemoimmunotherapy. Updated results from this trial of axi-cel as part of first-line therapy demonstrated a high rate of complete and durable responses, with >80% of patients remaining alive after at least 3 years of follow-up. Median OS was not reached, and the OS rate remained unchanged with additional follow-up of ≈4 years. Efficacy outcomes were favorable regardless of high-risk characteristics. No new safety signals were observed with long-term follow-up. As such, axi-cel may benefit patients with high-risk LBCL exposed to fewer prior therapies as a treatment option with curative intent.

Historically, R-CHOP (rituximab, cyclophosphamide, doxorubicin hydrochloride, vincristine sulfate, and prednisone) and other anti–CD20-targeted chemoimmunotherapies have had moderate first-line activity among a substantial proportion of patients with LBCL.¹ More recently, first-line polatuzumab vedotin combined with R-CHP (pola-R-CHP [rituximab, cyclophosphamide, doxorubicin hydrochloride {hydroxydaunorubicin}, and prednisone] [rituximab, cyclophosphamide, doxorubicin hydrochloride, and prednisone]) demonstrated superiority of 2-year estimated PFS over R-CHOP (77% vs 70%) in patients with IPI ≥2 in the POLARIX trial, although OS was similar between arms.¹² Estimated 3-year PFS with axi-cel (75%) was similar to the 2-year estimates of both arms in POLARIX in a higher-risk population. However, the CR rate in ZUMA-12 with axi-cel appeared higher than that observed with pola-R-CHP (86% vs 78%), although comparisons between these trials are limited by follow-up, eligibility criteria, and patient characteristics (including the proportion of patients with double- or triple-hit lymphomas). In addition to CAR T-cell therapy, several novel modalities used in the relapsed/refractory setting are undergoing assessment in the frontline.¹³ 

Notably, no patient relapsed after 18 months postinfusion in ZUMA-12. Remissions lasting beyond 2 years may be associated with durability, suggestive of axi-cel as a curative treatment option in the first-line setting. Indeed, the longer follow-up of trials of axi-cel in second and later lines has supported its use as therapy with curative intent.^8,14 Optimizing the utility of this treatment by administration in earlier lines may be associated with improved outcomes in LBCL. A recent analysis of axi-cel manufacturability in LBCL demonstrated an increase in first-pass manufacturing success and a greater proportion of naive-like T cells in leukapheresis material (previously demonstrated to be associated with improved outcomes) with second vs third or higher lines of therapy.^15,16 Additionally, a recent prospective real-world assessment of earlier CAR T-cell therapy in DLBCL showed improved T-cell fitness and a trend toward longer survival with second- vs third-line or higher therapy, which suggests that greater T-cell fitness might be contributing to the rate of durable complete remissions observed in this trial.¹⁷ Health economic assessment of second-line CAR T-cell therapy demonstrated earlier use of the curative therapy is cost-effective, stressing the importance of addressing delays or barriers to its access.^18,19 ZUMA-12 enrolled a small number of patients with relatively heterogeneous high-risk disease characteristics, and data on the benefit of first-line CAR T-cell therapy to date are limited. Further assessment of the impact of first-line CAR T-cell therapy on clinical and health economic outcomes is needed to determine its benefit.

Interim PET2 status may have the potential to inform timely treatment adjustment in the first line. A recent assessment of outcomes by interim PET2 status among 2 prospective cohorts with DLBCL treated with first-line standard of care showed interim PET2+ was associated with worse survival outcomes.³ Studies are ongoing to optimize the method for assessing interim PET2, although it has demonstrated utility as a prognostic indicator.²⁰ PET2+ has been shown to be a functional biomarker that suggests these patients are unlikely to benefit from continued chemotherapy. The GAINED study further demonstrated that consolidation with autologous stem cell transplantation among patients who were both interim PET2+ and PET4− may be associated with favorable outcomes; however, the patient population in ZUMA-12 represented those of higher risk.²¹ Patients in ZUMA-12, all with PET2+, had a high rate of ongoing CR (73%) at data cutoff, demonstrating that intervention with axi-cel in this population is associated with durable responses in the first line.

The long-term safety profile of axi-cel was consistent with previous clinical experience in later lines of therapy.^14,22,23 No new safety signals emerged with longer follow-up in this trial.⁹ Although prolonged cytopenias are common long-term adverse events among CD19-targeted cellular therapies, none lasted beyond the primary analysis data cutoff. Additionally, no secondary malignancies related to axi-cel, including those of T-cell origin, were observed in this trial. Comprehensive reporting of nonrelapse mortality is increasingly important as a measurement to identify long-term risks of death with CAR T-cell therapies, improving safety management after infusion. This is especially critical with first-line intervention with CAR T-cell therapy. A meta-analysis reported meaningful risk of nonrelapse mortality with CAR T-cell therapies in LBCL, driven primarily by infections and greater lines of therapy.²⁴ In ZUMA-12, nonrelapse mortality deaths were rare (n = 2), although population size may limit interpretation. Robust B-cell recovery was observed among patients in ongoing response at data cutoff. No significant differences in CAR T-cell expansion were observed when assessed by ongoing response, although the small number of patients who relapsed or did not respond limited this assessment.

In conclusion, results of the 3-year follow-up analysis of ZUMA-12 demonstrate the long-term durability of responses to axi-cel in high-risk LBCL as a part of a frontline regimen, supporting earlier use of this curative treatment option in these patients. Safety was consistent with previous clinical experience. A phase 3 randomized controlled trial has launched to examine the benefit of axi-cel compared with standard-of-care therapy as first-line treatment for patients with high-risk LBCL (ZUMA-23; NCT05605899).

The authors thank the patients who participated in this trial along with their families, caregivers, and friends; the trial investigators, coordinators, and health care staff at each study site; Charles Mercado, Possidonia Gontijo, Saran Vardhanabhuti, and Christine Lui at Kite, a Gilead Company, for support of statistical analysis; and Danielle Fanslow, of Nexus Global Group Science, for medical writing assistance, funded by Kite.

This study was funded by Kite.

Contribution: The study was designed in a collaboration between Kite (study sponsor) and authors. J.C.C., J. Wulff, C.M.W., J. Winters, I.K., H.X., and S.S.N. designed the study; J.C.C., M.D., J.M., M.L.U., C.T., O.O.O., A.F.H., C.S.U., Y.L., P.A.R., N.K., S.d.V., and S.S.N. provided study materials, enrolled and treated patients, and collated and assembled data; J.C.C., J. Wulff, C.M.W., J. Winters, I.K., H.X., and S.S.N. analyzed and interpreted data; and all authors participated in drafting the article, provided feedback throughout the development process, and approved the final submitted version.

Conflict-of-interest disclosure: J.C.C. reports a consulting/advisory role for ADC Therapeutics, Adicet Bio, AstraZeneca, Bristol Myers Squibb, Genentech, Genmab, Kite, a Gilead Company, and Novartis; speakers’ bureau participation for BeiGene and Eli Lilly; and research funding from ADC Therapeutics, AstraZeneca, Janssen, and Merck. M.D. reports honoraria from, a consulting/advisory role for, and speakers’ bureau participation for Gilead Sciences, Kite, Merck Sharp & Dohme (MSD), Novartis, AbbVie, and Roche; and research funding from Novartis, AbbVie, MSD, and Roche. J.M. reports honoraria from Curio, OncView, Physicians’ Education Resource, Seagen, and Targeted Oncology; consulting/advisory role for AbbVie/Pharmacyclics, ADC Therapeutics, Alexion, AstraZeneca, Aurobindo, Bayer, BeiGene, Celgene/Bristol Myers Squibb, Eli Lilly, Epizyme, Fosun Kite, Genentech/Roche, Genmab, Janssen, Karyopharm, Kite, Kyowa, MEI, MorphoSys/Incyte, Novartis, Pfizer, Seagen, TG Therapeutics, and Verastem; and research funding from Bayer, BeiGene, Celgene, Genentech, Gilead, Incyte, Janssen, Kite, Merck, Millennium, Novartis, Pharmacyclics, Portola, and Seagen. M.L.U. reports a consulting/advisory role for Gilead Sciences and Stemline. C.T. reports a consulting/advisory role for Amgen, Bristol Myers Squibb, Incyte, Kite, Novartis, Roche, and Takeda; research funding from Roche; and travel support from Bristol Myers Squibb, Incyte, Kite, Novartis, Roche, and Takeda. O.O.O. reports honoraria from Gilead Sciences and Pfizer; consulting/advisory role for AbbVie, ADC Therapeutics, Cargo Therapeutics, Caribou Biosciences, Epizyme, Gilead Sciences, Kite, Nektar, Novartis, Pfizer, and TGR; speakers’ bureau participation for ADC Therapeutics; and research funding from Allogene, Daiichi Sankyo, Kite, and Pfizer. A.F.H. reports a consulting/advisory role for Allogene Therapeutics, AbbVie, Adicet Bio, ADC Therapeutics, AstraZeneca, Bristol Myers Squibb, Caribou Biosciences, Genentech, Genmab, Karyopharm, Merck, Pfizer, Regeneron, Takeda, and Tubulis; and research funding from ADC Therapeutics, AstraZeneca, Bristol Myers Squibb, Genentech, Gilead Sciences, Kite, and Merck. C.S.U. reports a consulting/advisory role for Allogene, Ascentage, ADC Therapeutics, AbbVie, AstraZeneca, Atara, BeiGene, Epizyme, Genentech, Janssen, Loxo Oncology/Lilly, and Pharmacyclics; and research funding from AbbVie, AstraZeneca, Loxo Oncology, and Pharmacyclics. Y.L. reports a consulting/advisory role for Janssen, Sanofi, Neximmune, Caribou, Bristol Myers Squibb, Pfizer, and Regeneron; and research funding from Bristol Myers Squibb and Janssen. P.A.R. reports a consulting/advisory role for AbbVie, ADC Therapeutics, BeiGene, Bristol Myers Squibb, Genmab, Janssen, Genentech, Kite, Therapeutics, Pharmacyclics, and Sana Biotechnology; and research funding from Calibr, Celgene/Bristol Myers Squibb, AstraZeneca, Genentech, CRISPR Therapeutics, Fate Therapeutics, Kite, MorphoSys, Novartis, Tessa Therapeutics, and Xencor. N.K. reports honoraria from and a consulting/advisory role for Celgene, Gilead Sciences, and Novartis. S.d.V. reports data safety advisory board participation for BeiGene. J. Wulff reports employment with Kite. C.M.W. reports employment with and research funding from Kite; and stock or other ownership in Gilead Sciences. J. Winters reports employment with Kite. I.K. reports employment with Kite. H.X. reports employment with Kite, a Gilead Company. S.S.N. reports a consulting/advisory role for Adicet Bio, Athenex, Appia Bio, Allogene, Astellas Pharma, Athenex, Bluebird bio, Bristol Myers Squibb, Caribou Biosciences, Carsgen, Chimagen, Fosun Kite, Galapagos, GlaxoSmithKline, ImmunoACT, Incyte, Janssen, Jazz Pharmaceuticals, Kite, Merck, ModeX Therapeutics, MorphoSys, Orna Therapeutics, Sana Biotechnology, Sellas Life Sciences, Orna Therapeutics, GlaxoSmithKline, Synthekine, and Takeda; research funding from Adicet Bio, Allogene, Bristol Myers Squibb, Kite, Precision Biosciences, Cargo Therapeutics, and Sana Biotechnology; stock options from Longbow Immunotherapy; and patents, royalties, or other intellectual property related to cell therapy.

Correspondence: Julio C. Chavez, Moffitt Cancer Center, 12902 USF Magnolia Dr, Tampa, FL 33612; email: julio.c.chavez@moffitt.org.