AMELIA: A phase 1 trial of CAR T cells with dual CD19 and CD22 targeting in pediatric and young adult patients with R/R B-ALL

Chimeric antigen receptor (CAR) T cells that target CD19 or CD22 have demonstrated promising activity in relapsed or refractory (R/R) B-cell acute lymphoblastic leukemia (B-ALL), though as many as 42% of patients who initially respond to anti-CD19 CAR T-cell therapy experience CD19-negative relapse. Additionally, data from a phase 1 study of anti-CD22 CAR T-cell therapy suggests that escape by CD22 downregulation may explain the reduced CD22 antigen density seen in relapsed patients in that trial.¹

Cordoba, et al.¹ hypothesized that, as a single leukemic stem cell is unlikely to downregulate both of these targets at the same time, dual antigen targeting—via coadministration of two separate products, transduction of T cells with either two vectors, one bicistronic vector, or use of a tandem CAR—may prevent relapse. The investigators developed AUTO3, a CAR T-cell therapy with dual specificity for CD19 and CD22 encoded by a bicistronic vector. Here we report results of the phase 1 study of AUTO3.¹

Study design

This was a phase 1 preclinical and dose-escalation study of AUTO3 in pediatric and young adult patients with R/R B-ALL reporting pre-clinical experiments as well as safety and efficacy data. Patients were aged 1‒24 years and were enrolled in three UK centers. An adult cohort was planned but not initiated.

The primary endpoints of the study were:

• Incidence of Grade 3‒5 toxicity in the dose-limiting toxicity (DLT) period (30 days after AUTO3 infusion)
• Frequency of DLTs

Secondary endpoints included:

• The proportion of patients achieving morphological remission, defined as complete response (CR) or CR with incomplete hematologic recovery (CRi)
• Having a minimal residual disease negative response after AUTO3 infusion
• The frequency and severity of adverse events
• The expansion and persistence of AUTO3
• The duration of B cell aplasia
• Overall survival (OS) and event-free survival

Baseline characteristics

Overall, 23 patients were screened and 20 patients underwent leukapheresis; 19 of those 20 had CAR T-cell products generated, and 15 patients received an infusion of AUTO3. The median age was 8 years (range, 4–16 years). Baseline patient characteristics are shown in Table 1.

Table 1. Baseline characteristics at enrollment*

Results

Recognition of CD19 and CD22 by AUTO3 CAR T cells

AUTO3 was generated via transduction of autologous T cells with a bicistronic ɣ-retroviral vector encoding a CD19 and a CD22 CAR transcribed from a single promoter; the CARs were separated by a self-cleaving 2A peptide. The CD19 and CD22 binders were humanized in an attempt to reduce immunogenicity. The tumor necrosis factor receptor costimulatory domains for CD19 and CD22 were OX40 and 41BB, respectively, which have analogous immune functions. Functional testing of AUTO3 demonstrated effective killing of CD19 and CD22 expressing engineered cell lines.  

AUTO3 versus an FMC63-based CAR

AUTO3 was compared to a CD19 FMC63-based CAR used in tisagenlecleucel and demonstrated enhanced cytotoxicity against CD19 and CD22 expressing Raji cells in in vitro co-cultures (p = 0.0001). Unlike the FMC63-based cells, AUTO3 did not show progressive reduction in cytotoxicity. The superior efficacy of AUTO3 was maintained at lower doses of CAR T cells.

Phase 1 trial with AUTO3

Prior to lymphodepletion, 14 of the 15 patients had an evaluable morphologic bone marrow assessment; of those 14 patients, eight were in morphological relapse, and six had minimal residual disease-level disease. None of the 15 patients had concomitant extramedullary disease.

All 15 patients received a preconditioning regimen of fludarabine and cyclophosphamide for 2 days prior to AUTO3 infusion. Regarding AUTO3 dose:

• Four patients received 0.3‒2 × 10⁶ CAR T cells/kg
• Five patients received 3 × 10⁶ CAR T cells/kg
• Six patients received 4.3‒5 × 10⁶ CAR T cells/kg

AUTO3 infusion was infused as a split dose in patients with high disease burden (≥25% blasts) to mitigate the risk of severe toxicity.

Toleration of AUTO3

During dose escalation, nine of 15 patients (60%) experienced Grade 3‒5 toxicity; no DLTs were reported, and a maximum tolerated dose was not reached. Twelve patients (80%) developed cytokine release syndrome (CRS), which was mild in all cases, and two patients required tocilizumab (Table 2). No patients required intensive care due to CRS.

Table 2. CRS events per patient*

Four patients experienced neurotoxicity possibly related to AUTO3, which was mild in all cases (Table 3). One patient developed Grade 3 encephalopathy after a single dose of AUTO3; this was deemed more likely due to prior treatment with intrathecal methotrexate than AUTO3 infusion, given a complete absence of CRS symptoms and low levels of proinflammatory cytokines.

Table 3. Neurotoxicity events per patient*

Nine patients (60%) experienced Grade 3‒4 toxicity by Day 30, including:

• Fever (n = 7)
• Febrile neutropenia (n = 4)
• Neutropenia, anemia, and infections (all n = 2)
• Thrombocytopenia (n = 1)

The most common Grade 3‒4 events at any time after AUTO3 infusion were similar to those occurring in the first 30 days after infusion. There were no deaths due to adverse events.

Expansion and persistence of AUTO3

Substantial expansion of CAR T cells (maximum concentration >30,000 copies per µg DNA) was seen in ten patients (66%), with a median time to maximal expansion of 12 days (range, 7‒15 days). There were 13 patients who achieved CR/CRi, and those patients had AUTO3 expansion kinetics similar to what has been observed with tisagenlecleucel.

Patients who received an AUTO3 dose of 3 × 10⁶ cells/kg had a longer median duration of persistence (344 days; range, 63‒539 days) than those who received the lower (42 days; range, 20‒257 days), or higher dose (28 days; range, 19‒571 days). Patients in the higher dose cohort had low levels of CD4- and CD8-naïve cells, which may have affected the persistence of AUTO3 in this cohort.

Relapses and lack of CAR-T persistence

As noted previously, 13 of 15 patients (86%) achieved CR/CRi at 1 month after AUTO3 infusion, and the rate of complete molecular remission was 80% at 1 month and 86% at 2 months.

• Two patients did not respond to treatment
• Nine of the 13 patients in CR have relapsed
  • CD19 loss was detected in blasts in three patients at relapse
• Three patients in CR have received new treatment for ALL while in remission
• One patient maintained complete molecular response without any additional treatment at 18 months

At the cut-off date six patients (40%) are alive:

• The OS rate was 80% at 6 months and 60% at 12 months for all treated patients.
• The OS rate was 90% at 6 months and 70% at 12 months for CAR-naïve patients who received ≥3 × 10⁶ cells/kg AUTO3.
• The event-free survival rates at 6 months and 12 months were 48% and 32% for all treated patients and 57% and 46% for CAR-naïve patients, respectively.

Conclusion

This phase 1 study demonstrates the feasibility of bicistronic CAR T-cell therapy targeting both CD19 and CD22 in pediatric and young adult patients with R/R B-ALL. There was no evidence of increased toxicity with the dual target therapy, even in patients with high tumor burden. Patients with high tumor burden received a split dose; however, despite similar safety profiles, the small sample sizes of the single-dose and split-dose groups makes comparison difficult. In total, 13 of the 15 patients achieved CR/CRi, though nine of those 13 relapsed, and data suggest that lack of long-term persistence was the likely cause of relapse. Additional strategies are required to improve the persistence of AUTO3 before a true assessment of its impact on relapse rates can be determined.