Relmacabtagene autoleucel (relma‐cel) CD19 CAR‐T therapy for adults with heavily pretreated relapsed/refractory large B‐cell lymphoma in China

Abstract Background Despite numerous chimeric antigen receptor T‐cell (CAR‐T) trials conducted in China, no CAR‐T has been registered in the country. Furthermore, China law and regulations restrict the export of patient material for CAR‐T manufacture abroad. Relma‐cel (JWCAR029), an anti‐CD19 product produced with a commercial‐ready process in China, was evaluated in the first prospective, single‐arm, multicenter, pivotal study of CAR‐T therapy conducted under Chinese IND to support an NMPA‐accepted BLA submission in relapsed/refractory (r/r) LBCL (NCT04089215). Methods Patients were randomized to receive either 100 × 106 (low dose, n = 27) or 150 × 106 (high dose, n = 32) CAR+ T‐cells as a single infusion following lymphodepleting chemotherapy (fludarabine 25 mg/m2 and cyclophosphamide 250 mg/m2 daily × 3), and then, monitored for efficacy and safety outcomes and pharmacokinetics. The primary endpoint was ORR at 3 months, as assessed by the investigators. Secondary endpoints included DOR, PFS, OS, and adverse event frequency/severity and cell expansion kinetics. Results As of the data cutoff on 17 June 2020, 68 patients were enrolled, and 59 were treated. Among the 58 efficacy‐evaluable patients, the primary endpoint of 3 month ORR was 60.3% (95% CI, 46.6–73.0), excluding the null hypothesis rate of 20%. Any grade and severe grade CRS occurred in 47.5% and 5.1%, respectively, and any grade and severe grade neurotoxicity events occurred in 20.3% and 5.1%. Conclusions Relma‐cel met the primary endpoint analysis and demonstrated a high rate of durable responses and low rate of CAR‐T‐associated toxicities in patients with r/r LBCL in a multicenter trial supporting regulatory submission in China.


| INTRODUCTION
Non-Hodgkin's lymphoma (NHL) was diagnosed in over a half million persons worldwide in 2018. 1 In China, 68,500 new cases, 37,600 deaths and 237,000 existing cases of NHL were estimated in 2016. 2 Chinese large B-cell lymphoma (LBCL) patients typically receive R-CHOP as frontline therapy. For those with relapsed or refractory disease, salvage chemotherapy regimens followed by high-dose chemotherapy and autologous stem-cell transplant (ASCT) is a common strategy, however, only half of those with chemo-responsive disease receive ASCT in China. 3,4 Once failing second-line therapy, these patients have few therapeutic options beyond additional cycles of salvage chemotherapy, which has generally been associated with low response rates and limited overall survival. 5,6 Commercially available CAR-T cell products targeting CD19 have been available in the United States (US) since 2017 and in European Union (EU) since 2018 for patients with r/r DLBCL after failing two prior lines of therapy. These products, including axicabtagene ciloleucel and tisagenlecleucel, have demonstrated high rates of durable disease response in registrational clinical trials. 7,8 However, neither of these products, nor any other CAR-T product has been approved for commercial use for r/r LBCL in China, despite the country having the second largest number of CAR-T trials after the US. This is due to two primary factors: (a) Most CAR-T trials in China are single-center studies not conducted under a Clinical Trials Application (CTA) with National Medical Products Administration (NMPA), and thus, the data are not able to support product registration, and (b) unlike the EU, national laws and regulations prohibit the export of patient material, such as an apheresed peripheral blood mononuclear cells, that would enable access to commercial CAR-T manufacturing outside the country to provide early patient earlier access to these agents.
Relma-cel is a CD19-targeted, second generation CAR-T cell product with a 4-1BB costimulatory domain manufactured in China. In fact, relma-cel expresses the same CAR as another CAR-T manufactured in the US, lisocabtagene maraleucel, which has demonstrated high response rates in r/r LBCL with low levels of CAR-T-associated toxicity. [9][10][11] However, relma-cel uses a commercial-ready process developed in China that does not require separate CD4 and CD8 T-cell production trains to provide a wide range of doses with consistent product attributes. 12 A phase I study of relma-cel (NCT03344367 and NCT03355859) demonstrated preliminary safety and efficacy in r/r NHL patients over a dosing range from 25 million to 150 million CAR+ T-cells. [13][14][15] These data supported the design and execution of the first prospective, single-arm, multicenter, registrational study of CAR-T therapy conducted under Chinese Investigational New Drug (IND) to enable an NMPA-accepted Biologic License Application (BLA) submission in r/r LBCL (NCT04089215). Herein, the results of this pivotal trial in China are reported.

| Study design
This was a single-arm, open-label, multicenter clinical trial design evaluating relma-cel in adult (≥18 years) patients with CD19+ r/r LBCL, study schema provided in Figure 1. This study was approved by Ethics Committee at each site and conducted in accordance with Good Clinical Practice guidelines of the International Committee on Harmonization. After providing informed consent, patients were randomized to receive either 100 × 10 6 (low dose, n = 27) or 150 × 10 6 (high dose, n = 32) CAR+ T-cells. Leukapheresed patients could receive bridging chemotherapy based on investigator assessment of Secondary endpoints included DOR, PFS, OS, and adverse event frequency/severity and cell expansion kinetics. Results: As of the data cutoff on 17 June 2020, 68 patients were enrolled, and 59 were treated. Among the 58 efficacy-evaluable patients, the primary endpoint of 3 month ORR was 60.3% (95% CI, 46.6-73.0), excluding the null hypothesis rate of 20%. Any grade and severe grade CRS occurred in 47.5% and 5.1%, respectively, and any grade and severe grade neurotoxicity events occurred in 20.3% and 5.1%. Conclusions: Relma-cel met the primary endpoint analysis and demonstrated a high rate of durable responses and low rate of CAR-T-associated toxicities in patients with r/r LBCL in a multicenter trial supporting regulatory submission in China. disease burden and tumor progression risk. Lymphodepleting (LD) chemotherapy was administered over 3 days, including fludarabine (25 mg/m 2 i.v. daily) and cyclophosphamide (250 mg/m 2 i.v. daily), followed 2-7 days later by a single intravenous infusion of relma-cel at the assigned dose. Efficacy was assessed at baseline, 1, 3, 9, 12, 18, and 24 months post infusion. Individuals receiving bridging chemotherapy were restaged prior to beginning LD chemotherapy. Safety events were monitored from beginning of LD through 2 years follow-up.

| Participant population
Eligible adult patients had measurable, histologically confirmed r/r LBCL, including several LBCL subtypes (nonspecific DLBCL [DLBCL-NOS], high-grade B-cell lymphoma with MYC and BCL2, and/or BCL6 rearrangements [double-/triple-hit lymphoma, DHL/THL], primary mediastinal Large B-cell lymphoma [PMBCL], grade 3b follicular lymphoma [FL], and FL-transformed DLBCL) on the basis of the 2008 WHO Classification, 16 and not had allogeneic transplant within 90 days, or primary central nervous system (CNS) lymphoma. Refractory or relapsed patients must have previously been treated with at least two lines of therapy containing an anti-CD20 monoclonal antibody and anthracycline-containing regimen. Those receiving ASCT must have relapsed, progressed disease or never achieved remission within 24 months posttransplant.

| CAR construct and CAR-T cell manufacturing
Relma-cel is comprised of autologous CD4+ and CD8+ T-cells that have been transduced to express a CD19-specific chimeric antigen receptor (CAR) and a truncated epidermal growth factor receptor (EGFRt). The CAR receptor is comprised of an extracellular scFv-binding domain derived from a murine CD19-specific hybridoma cell line (FMC63), fused to the 4-1BB and CD3 ζ chain signaling domains. Autologous mononuclear cells (MNCs) are collected from patient via leukapheresis. The T-cells are specifically enriched from apheresis using CD4 and CD8 microbeads. The selected T-cells are activated using CD3/CD28 microbeads. The activated T-cells are transduced ex vivo with nonreplicative self-inactivating (SIN) lentiviral vector containing the CAR transgene. The transduced T-cells are expanded in cell culture. The expanded CAR-T cells are harvested and formulated into a cryopreservation medium. The drug product consists of a single frozen cell suspension formulated in a cryopreservation medium.

| Outcomes
The primary endpoint was the proportion of patients achieving objective response at 3 month by investigator assessment, using the Lugano Classification. 17  Enzyme-linked Immunosorbent Assay [ELISA]), serum antitherapeutic antibodies (ATA; by MSD), and CAR-T cell concentrations in the peripheral blood (PB) using transgene copy number per microgram DNA (by qPCR) and CD3+ CAR+ cell frequency (by flow cytometry). ATA testing uses antibody targeting the scFV part of CAR Extracellular domain.

| Statistical analysis
The primary analysis was conducted based on the data cutoff on 17 June 2020 with 58 patients assessed for 3 month objective response rate (ORR) in the efficacy analysis set. Under the null hypothesis assumption of ORR ≤20% for standard therapy in this setting and an alternative hypothesis assumption of 40% ORR, with the sample size of 58 patients, this study has ≥92% power and one-sided α of 0.025. DOR, PFS, and OS were summarized using Kaplan-Meier methods. 18 DOR was defined as time of first response to progression or death. PFS was defined as the time from infusion to disease progression or death and OS was defined as the time from the infusion to death. Patients were censored either at the relma-cel infusion date, if no radiologic assessment post infusion, or the date of last tumor assessment prior to starting new treatment before PD, if follow-up monitoring was discontinued beyond PD or death, or if death or PD occurred after ≥2 consecutive missed restaging visits.
Pharmacokinetics (PK) parameters were calculated using non-compartmental analyses with Phoenix WinNonlin 8.0 (Certara) for Cmax (observed maximum plasma concentration following infusion), Tmax (time at which maximum plasma concentration was observed), and AUC 1-29 (exposure calculated as the product of plasma concentration and time using a trapezoid rule). Boxplots of PK parameters analyzed by age, sex, ATA positive or negative post infusion, and tocilizumab or steroid use; nonresponse, OR or CR; and CRS or neurotoxicity (NT) by severity grade were calculated using Wilcoxon test.

| Patient characteristics
Between Nov, 2017 and Dec, 2019, 90 patients were screened, 68 apheresed, and 59 treated ( Figure 2). Six patients discontinued prior to relma-cel infusion due to disease progression (one death and five withdrawal) and three patients were at dose 25 × 10 6 or 50 × 10 6 CAR+ T-cells. Cell product was manufactured for all apheresed patients, median production time was 19 days (range, . Treated patients had a median age of 56.0 years (range, 18-75); included several subtypes including 41 (69.5%) with Nonspecific diffuse large B-cell lymphoma (DLBCL-NOS), and six (10.2%) had previously undergone ASCT ( Table 1). Sum of perpendicular diameters (SPD) ≥5000 mm 2 was present in 14 (23.7%) of 59 patients; 26 (44.2%) in 59 patients had a bridging chemotherapy and 23 (47.9%) in 48 patients had a high-risk International Prognostic Index (IPI) score ≥3 ( Table 1). The median doses infused in low-dose group and high-dose group F I G U R E 2 Patients screening, enrollment and treatment. Ninety patients were screened, 68 apheresed, and 59 treated. Among the six patients who were not received relma-cel, one died before infusion, five withdraw. *One patient, excluded from efficacy analysis set, received relma-cel infusion that did not meet potency threshold criterion, but achieved CR by D29 that continued for more than 1 year.  were 99.7 × 10 6 (range, 80.1-101.3) and 150.0 × 10 6 (range, 120.0-156.4) CAR+ T-cells, respectively. No meaningful differences in baseline characteristics were apparent between dose groups (Table 1).

| Efficacy
Fifty-eight patients were evaluable for efficacy outcomes, and the primary endpoint of 3 month ORR was 60.3% (95% CI, 46.6-73.0). The null hypothesis of a 3 month ORR ≤20% was rejected (p < 0.001). The excluded patient received product that did not meet potency threshold criterion, but achieved CR at D29 that continued for >1 year. As of data cutoff date, BOR rate was 75.9% (95% CI, 62.8-86.1) with a 51.7% CR rate (95% CI, 38.2-65.1; Figure 3A; Table S4). A sensitivity analysis was conducted using response restaging adjudicated by a prospectively established independent review committee (IRC) that yielded point estimates concordant with investigator assessment for both the primary endpoint and BOR rates (Table S3). High rates of response were observed across all key covariates, including age, SPD, IPI score at enrollment, ATA positive or negative post infusion, cell-of-origin subtype, Eastern Cooperative Oncology Group (ECOG), and use of tocilizumab or glucocorticoids ( Figure 3B). With median follow-up of 8.9 months, median OS was not reached (95% CI, 10.3 to NA). Median PFS and DOR were 7.0 months (95% CI, 5.5 to NA) and 8.0 months (95% CI, 6.0 to NA), respectively, however, more than half of patients were censored prior to the median OS follow-up ( Figure 3C). In efficacy analysis set (58 patients), the median time to OR and CR evaluated by investigator were 0.92 and 0.95 month, respectively (Tables S6 and S7). Of 55 patients with baseline and at least one posttreatment imaging evaluation, the median best percentage change of SPD with treatment was −74.7% (range, −100-136.4%), with 42 (76.4%) patients having >50% decrease from baseline ( Figure 3D).

| Safety
Any grade adverse events (AEs) occurring in ≥10% of the 59 patients treated are shown in Table 2. Twenty-eight patients (47.5%) experienced CRS of any grade. Grade 3 and 4 CRS F I G U R E 3 Efficacy assessed by investigators. A, BOR evaluation among the 58 patients who received relma-cel infusion, an anti-CD19 chimeric antigen receptor T-cell therapy (ORR defined as CR plus PR). B, Forest plot showing the result of ORRs. CI=confidence interval. C, Kaplan-Meier curve of DOR among 44 patients, PFS and OS in the 58 patients. DOR, Time from the date of first documented disease response (CR or PR) to the date of first documented PD or death due to LBCL. PFS, Time from the date of relma-cel infusion to the date of PD or death from 58 patients. OS, time from the date of relma-cel infusion to the date of death. D, The change of tumor volume from the baseline according to SPD in 55 evaluable cases. The other four cases were excluded because they had clinical disease progression prior to the 1 month restaging assessments. The vertical axis indicates best percent changes from the baseline after relma-cel infusion for 6 month. Patients are indicated on the horizontal axis. Negative values in the vertical axis indicate tumor volume decrease from the baseline. NA, Not Available.  was observed in two patients (3.4%) and one patient (1.7%), respectively (Tables 2 and 3). The median onset of CRS was at 4.5 days (range, 1-10) after infusion with a median duration of 7.0 days (range, 1-118; Table S5). Neurologic events occurred in 12 patients, with only 3 (5.1%) having severe grade events (all Grade 3; Tables 2 and 3). Median onset of NT was  Table S5). CRS and NT were generally manageable and all cases resolved except one patient who died at day 8 due to sepsis with ongoing grade 4 CRS, but not related to CAR-T cell therapy as evaluated by investigator. Sixteen (27%) patients received tocilizumab and six (10%) corticosteroids for CRS, and one (1.7%) patient received tocilizumab and three (5.1%) corticosteroids for NT in accordance with the protocol-defined toxicity management algorithm 19 (Table 3).

| Cellular kinetics
For 58 efficacy-evaluable patients, median Cmax was 25333.5 copies/µg DNA by qPCR and 23.7 CD3+CAR+ cells/μL by flow cytometry with a median Tmax of 8.5 days and a median AUC 1-29 of 249744.8 day*copies/µg ( Figure 4A; Tables S1 and S2). Correlations between PK parameters (Cmax, Tmax, and AUC) and selected clinical covariates (age, sex, ATA positive or negative post infusion, and corticosteroid or tocilizumab use) demonstrated no potential association except that higher Cmax and AUC were observed in those with tocilizumab use ( Figure 4B; Figure S2). This is thought to be because patients with greater Cmax and AUC are more likely to develop sever CRS/NT and are therefore more likely to require tocilizumab therapy. Unlike a traditional drug, no relationship was detected between CAR-T cell kinetics parameters and response ( Figure S5). Although the relationship is generally expected for most drugs, the lack of a relationship here may be a result of the capacity of CAR-T to proliferate in vivo and heterogeneity across a patient's product, native immune system, and tumor burden. Therefore, more studies are needed to confirm the hypothesis. CD8+ and CD4+ CAR+ T-cells showed similar PK profiles over time post infusion ( Figure 4D) with both T-cell subsets peaking around day 8 and the mean values of CD8+ T-cells in the blood exceeding CD4+ T-cells in the blood at each time point throughout the time course. Mean peak concentrations by flow for low-and high-dose groups were dose proportional for CD4+ T-cells, but were more than dose proportional for CD8+ T-cells ( Figure S1). No difference in exposure was observed between two dose groups. Higher Cmax for CAR+ T-cells were observed in patients with any grade neurologic events and any grade CRS events and AUC 1-29 values for sever CRS and any grade NT ( Figure 5A,B).

| Serum biomarkers
Patients who achieved any response had lower baseline levels of serum TNF-α and responding patients also had lower peak concentrations for TNF-α and IFN-γ (p < 0.01; Figure 4C). No association was observed between baseline levels, peak concentrations or the magnitude of change for other cytokine levels tested and disease response ( Figure S3). Most peak serum cytokines (TNF-α, MCP-1, IL-15, IL-2, IFN-γ, CRP, and ferritin) showed higher in those with serve CRS events, however, TGF-β1 was lower in serve CRS. IL-8 had no relationship in those with any grade CRS/NT ( Figure S4). Peak serum IL-6 levels were higher in those with any grade neurologic events and any grade CRS events, while IL-15 was higher in serve CRS/NT ( Figure 5C).

| Anti-treatment antibodies (ATA)
No pre-infusion serum ATA was detected in any patient; post infusion, elevated serum ATA was observed in 10 (16.9%) patients at 6 months, among them five patients had complete remission (CR), three partial remission (PR), and two progressive disease (PD) at 6 months (data not shown).

| DISCUSSION
This is the first prospective, single-arm, multicenter, pivotal study of a CD19-specific CAR-T conducted under Chinese IND to support an NMPA-accepted BLA submission in patients with r/r LBCL. In this study, relma-cel demonstrated nearly 76% ORR and nearly 52% CRR in a cohort of r/r LBCL patients with many poor risk features, including 23.7% with high volume disease (SPD ≥5000 mm 2 ), 44.2% requiring bridging chemotherapy for disease control, and 39.0% with IPI score ≥3. Many of these responses were durable given the current study follow-up, with median DOR of 8 months with a plateau at approximately 48%. Toxicities commonly associated with anti-CD19 CAR-T were relatively low with any grade and severe CRS in 47.5% and 5.1%, respectively, and any grade and severe neurotoxicity in 20.3% and 5.1%, respectively. These low rates were obtained by using a protocol-defined toxicity management algorithm, did not result from significant tocilizumab or corticosteroid use, which was required in only 28.8% and 15.2% of treated patients, respectively. Pharmacokinetic parameters showed similar expansion and contraction kinetics over time as has been reported by several groups. [20][21][22][23][24] F I G U R E 5 Analysis of serum biomarkers (CAR-T expansion and Cytokines) associated with CRS/NT classified as none, grade 1-2 and grade ≥3 (**p < 0.01, ***p < 0.001). A, The comparison of AUC 1-29 and CRS/NT. B, The comparison of Cmax and CRS/NT. C, comparison of peak measured serum IL-6 levels and CRS/NT. CRS, cytokine release syndrome; NT, neurotoxicity, IL-6, Interleukin-6. p-values were calculated by means of the Wilcoxon rank-sum test.