CD19/CD22双价CAR-T细胞治疗儿童、青少年及年轻成人B-ALL：1期试验最终结果

Multiantigen targeting chimeric antigen receptor (CAR) T cells have emerged as a strategy to mitigate antigen escape observed after single antigen targeting therapy. Our initial experience with a bivalent CD19.22.BBζ CAR T-cell construct in children, adolescents and young adults (CAYA) with B-cell acute lymphoblastic leukemia (B-ALL) demonstrated limitations in CD22 recognition, but a tolerable safety profile and efficacy prompted further evaluation. This trial enrolled patients between the ages of 3–39 with relapsed/refractory B-ALL. Following dose-escalation, patients who enrolled at the recommended phase 2 dose (RP2D) of 3×106 transduced CAR T cells/kg constitute this report. 30 CAYA were treated at the RP2D; 28 with B-ALL and 2 with Burkitt lymphoma. Across patients with B-ALL, 20 (71.4%) patients developed cytokine release syndrome (CRS); only 2 (10%) were grade >3. Grade 3 immune effector cell-associated neurotoxicity syndrome (ICANS) occurred in 3 (10.7%) patients; there were no cases of immune effector cell-associated hemophagocytic lymphohistiocytosis-like syndrome. Following a protocol amendment to evaluate the efficacy of siltuximab as a first-line treatment of CRS, one patient received siltuximab with full resolution of CRS after two doses without needing additional anti-cytokine-directed therapies. A measurable residual disease-negative complete remission (CR) was attained in 25 (89.3%) patients, including 6 who had neither CRS nor ICANS. 23 patients (82.1%) proceeded directly to hematopoietic stem cell transplant (HSCT) following CAR T-cell infusion within a median of 51 days (range, 45–68 days), supporting the utility of this construct as a bridge to HSCT. All three non-responders had persistent non-central nervous system (CNS) extramedullary disease (EMD), although four of seven patients with non-CNS EMD achieved a CR. Median relapse-free survival among the 25 patients achieving CR was not reached, and the median overall survival for all 28 patients was 34 months (95% CI 17 to not estimable) from infusion. This extended experience demonstrates that CD19.22.BBζ CAR T-cell therapy is safe and clinically active, particularly as a bridge to HSCT. Non-response was confined to patients with non-CNS EMD, highlighting the persistent challenge of effectively targeting EMD and informing the design of future CAR constructs.

Trial registration number

NCT03448393.

While CD19-targeted chimeric antigen receptor (CAR) T cells have revolutionized outcomes for children, adolescents, and young adults (CAYA) with relapsed/refractory B-cell acute lymphoblastic leukemia (B-ALL), loss or downregulation of the target antigen has emerged as a mechanism of resistance. With a goal of enhancing durable remissions through multiantigen targeting, we tested a novel bivalent CD19.22.BBζ construct in a phase 1 dose-escalation study (NCT03448393). The initial experience with 20 CAYA with B-ALL and one with Burkitt lymphoma (BL) across three dose levels revealed a tolerable safety profile with dose-dependent response; the overall complete response (CR) rate was 60%, with a CR rate of 83.3% at dose level 3 (DL3), 3×106 transduced CAR T cells/kg, which was deemed the recommended phase 2 dose (RP2D).

Despite limited persistence and suboptimal CD22 targeting in vivo, given the tolerability and efficacy signal, additional patients were treated at the RP2D to explore new aims. Eligibility was expanded to include patients with isolated central nervous system (CNS) and CNS3 disease, and we prospectively tested siltuximab as first-line treatment of cytokine release syndrome (CRS). Incorporating additional correlative biology, we now report on the complete cohort treated at the RP2D.

This was a single-center, phase 1 dose-escalation study of CD19.22.BBζ CAR T cells in CAYA with CD19+/CD22+ B cell malignancies conducted by the Pediatric Oncology Branch, National Cancer Institute. This report focuses on all patients with B-ALL treated at the RP2D at DL3, 3×106 transduced CAR T cells/kg (including those treated during dose-escalation). Data cut-off was February 15, 2025.

Prior to dose-expansion, the protocol was amended to stratify patients into cohorts based on whether they were CAR-naïve (inclusive of patients who received an interim hematopoietic stem cell transplant (HSCT) following prior CAR) or CAR-pretreated. However, given limited accrual in the CAR-pretreated cohort, all patients were analyzed collectively. A later amendment added one-third cohort with isolated CNS (iCNS) disease and those with CNS3 involvement.

Participants received one of two lymphodepleting chemotherapy (LD) regimens: fludarabine (flu) 25 mg/m2 days −4 to −2 and cyclophosphamide (Cy) 900 mg/m2 on day −2 for CAR-naïve patients, or intensified LD with influenza 30 mg/m2 days −5 to −2, and Cy, 600 mg/m2 on days −3 and −2 for CAR-pretreated patients and re-infusions to overcome concerns for immune mediated rejection (online supplemental Appendix).

CAR T cells were manufactured using CliniMACS Prodigy with a CD19.22.BBζ bivalent vector incorporating the FMC63 murine single chain variable fragment (scFv) for (anti-CD19) and m971 human scFv (anti-CD22) as previously described.

CRS, immune effector cell-associated neurotoxicity syndrome (ICANS), and immune effector cell-associated hemophagocytic lymphohistiocytosis-like syndrome (IEC-HS) were graded using the American Society for Transplantation and Cellular Therapy consensus criteria. Neutropenia and thrombocytopenia were graded using immune effector cell-associated hematotoxicity (N-ICAHT and T-ICAHT) scoring. Additional adverse events were graded using Common Terminology Criteria for Adverse Events V.5.0.

The general approach to CRS management used tocilizumab ± steroids to prevent grade >3 CRS. With an amendment at the RP2D, use of siltuximab (a direct interleukin (IL)-6 antagonist) as first-line treatment for CRS was implemented to explore its efficacy in toxicity management.

Disease assessment and response evaluation, including measurable residual disease (MRD) by multiparametric flow cytometry (MFC) and next generation sequencing (NGS) with clonoSEQ (Adaptive Biotechnologies) when available, was performed as described in the online supplemental Methods. Patients were classified as having low-disease burden if bone marrow (BM) was <5% blasts (M1), whereas high-disease burden was classified as ≥5% blasts (M2: 5–25%, M3: ≥25%). B-cell aplasia was defined as CD19+B cells <10/mcL in the peripheral blood (PB).

CAR T-cell expansion was analyzed by MFC in the PB, BM and cerebral spinal fluid (CSF) at protocol-specified and/or clinically relevant time points. Circulating CAR T cells/mcL were calculated based on concurrent absolute lymphocyte counts, whereas BM and CSF quantification was reported as % CAR+ T cells. Plasma biomarkers were evaluated at four-unique time points (day 0, day+5, day+9, day+14) using the Olink Target-48 Cytokine panel per the manufacturer’s standard protocol (Uppsala, Sweden). Samples were run in a single batch and Olink NPX values reported as pg/mL based on a standardized four-parameter logistic curve. A select number of CAR T-cell infusion products were analyzed for vector copy number, T-cell phenotyping by flow cytometry, metabolic characterization by Seahorse assay, and gene expression analysis by both NanoString nCounter and CITEseq (online supplemental Methods).

Statistical analyses were primarily descriptive and performed using GraphPad Prism V.10.6.1 and RStudio V.2025.09.1+401. All analyses were performed using non-parametric tests with p values <0.05 as the threshold of statistical significance. Comparisons were done using Mann-Whitney or Wilcoxon tests. Relapse-free survival (RFS) was defined as the time from initial CR (day 28 restaging) to relapse, death from any cause or date of last follow-up, restricted to patients who achieved remission. Overall survival (OS) was defined as the time from CD19.22.BBζ infusion to death from any cause. Kaplan-Meier survival curves were generated using R packages survminer and ggsurvplot.

Across 39 patients enrolled, 8 were treated at dose level (DL)1 and DL2; 1 did not proceed to infusion due to progressive disease (PD). The remaining 30 patients were treated at DL3 (3×106 transduced CAR T cells/kg), the RP2D, including 2 with BL (online supplemental Results) and 28 with B-ALL, 12 of whom were previously reported, and 16 newly reported in this analysis (figure 1A).

Cohort data: (A) consort diagram demonstrating all enrolled patients and further broken down into dose escalation and expansion cohorts, with further description of patients treated at dose level 3. ˆOne patient not infused due to progressive leukemia and concurrent infection. *12 patients with B-ALL at DL3 previously reported, **16 patients with B-ALL at DL3 newly reported; (B) cytogenetics of patients with B-ALL treated at dose level 3; n=28 patients, some of whom had more than one aberration. B-ALL, B-cell acute lymphoblastic leukemia; CAR, chimeric antigen receptor; DL, dose level; LD, lymphodepleting chemotherapy.

Among 28 patients with B-ALL, the median age at CAR T-cell infusion was 21.3 years (range, 6.3–38.0 years) (table 1). Approximately half were male (n=15, 53.6%), with equal numbers of Hispanic and non-Hispanic patients (n=14 each, 50%). The median number of prior therapies (excluding HSCT) was 4 (range, 2–9). Prior therapy included HSCT in 12 (42.9%), blinatumomab in 14 (50%), inotuzumab in 9 (32.1%), and CD19 or CD19/22 CAR T cells in 5 (17.9%). Most (26 (92.9%)) were CAR-naïve (including those with post-CAR HSCT, n=3) and 2 were CAR-pretreated. Baseline cytogenetics were variable (figure 1B).

Demographics of patients with B-ALL

All had detectable disease pre-infusion; 14 (50%) M1, 3 (10.7%) M2, 9 (32.1%) M3, and 2 (7.1%) with isolated EMD: 1 with iCNS disease and 1 with isolated non-CNS EMD. All patients were CNS1 by cytology and CSF counts; however, four had low-level CSF blasts detected by MFC. Seven (25%) had non-CNS EMD (online supplemental Table 1).

20 (71.4%) patients developed CRS at a median of 4 days (range, 0–17) (table 2). In 18 (90%) this was grade 1–2; 2 (10%) experienced grade 3 CRS. CRS resolved without therapeutic intervention in 13 patients, whereas 6 required both tocilizumab and steroids and 1 required tocilizumab alone. Two additional patients with CRS did subsequently receive steroids for ICANS but not for CRS. Following the amendment to evaluate the efficacy of siltuximab, only one of six who were treated developed CRS necessitating therapeutic intervention. In this case, two doses of siltuximab alone effectively treated grade 3 CRS.

Toxicities and outcomes

Three patients (10.7%) developed ICANS, all grade 3. One case constituted a dose-limiting toxicity during dose-escalation necessitating treatment of three additional patients prior to establishing the RP2D. Clinical manifestations of ICANS included right-sided weakness, disorientation, and seizure. All three received intrathecal hydrocortisone and systemic dexamethasone and recovered within 48 hours.

20 patients (71.4%) developed N-ICAHT, of which, 11 (55%) were ≥grade 3. The onset of severe neutropenia (absolute neutrophil count ≤500) occurred at a median of 5 days post-infusion (range, 0–17) and lasted for a median of 13 cumulative days (range, 2–30). 10 (35.7%) met criteria for early T-ICAHT with a median of 18.5 cumulative days of thrombocytopenia (platelet count <50/mcL) (range, 1–30 days). Four (14.3%) developed an infection (<grade 3) within 30 days post-infusion. No patient developed IEC-HS. All experienced at least one grade >3 adverse events (online supplemental Table 2).

At approximately 1-month post-infusion, 25 patients (89.3%) achieved an MRD-negative CR with no evidence of EMD. Of these responders, 18 had concurrent BM NGS, 10 (55.6%) of whom were NGS-negative. All 8 (44.4%) patients with positive NGS went to consolidative HSCT. By prior CAR T-cell exposure, 23/25 (92%) CAR-naïve patients and 2/2 (100%) CAR pretreated patients achieved CR. The three non-responders had baseline non-CNS EMD with variable post-infusion response: one had stable disease in the BM and EMD, one achieved MRD-negativity in the BM with persistent EMD, and one achieved MRD-negativity of the BM but progressed with EMD manifestations of lineage switch (LS) to acute myelogenous leukemia. Collectively, 4/7 (57.1%) with baseline non-CNS EMD had full resolution.

23 responders (92%) proceeded directly to consolidative HSCT following CD19.22.BBζ infusion within a median of 51 days (range, 45–68 days). Among these, 14 (60.9%) underwent their first HSCT and 9 (39.1%) a second HSCT. Of the 2 responders that did not proceed to HSCT, one relapsed within 5 months and underwent CD19.22.BBζ reinfusion but ultimately died of PD, while the other relapsed after 33 months and pursued HSCT following salvage therapy. In total, 8 responders relapsed at a median of 10 months (range, 3–34 months) following CAR infusion, including 6 with post-HSCT relapse at a median of 11 months post-HSCT (range, 2–12 months) (figure 2A). Among 8 patients who relapsed, phenotypic data at the time of relapse was only available for 4 patients: one had CD19-negative disease (although after blinatumomab following HSCT) and one had CD22-negative disease.

Survival, outcomes and correlatives: (A) swimmers plot demonstrating the longer-term outcomes of the 25 patients with B-ALL who had a CR; Kaplan-Meier analysis of, (B) relapse-free survival from the time of CAR T-cell induced CR in 25 patients with B-ALL treated at the recommended phase 2 dose who achieved CR and (C) overall survival from the time of CAR T-cell infusion in all 28 patients with B-ALL treated at the recommended phase 2 dose. (D) Cytokine analysis in patients with and without CRS as analyzed using the Olink Target-48 Cytokine panel. Cytokines with a log2 fold change of greater than 1 or less than −1 and an adjusted p values<0.05 are highlighted. B-ALL, B-cell acute lymphoblastic leukemia; CAR, chimeric antigen receptor; CR, complete response; CRS, cytokine release syndrome; CXCL, chemokine (C-X-C motif) ligand; lHSCT, hematopoietic stem cell transplant; IFN, interferon; IL, interleukin; MMP1, matrix metalloproteinase-1.

16 (57.1%) patients (all who achieved MRD-negative CR) remain alive at a median of 23.5 months (range, 6–60 months), with 15 in ongoing CR following consolidative HSCT. Across 12 patients that died, 8 died of PD, 2 of HSCT-related mortality, 1 from progressive LS, and 1 had an unknown cause of death. The median RFS for the 25 patients with MRD-negative CR was not reached; the median OS for all 28 patients was 34 months (95% CI 17 to not-achieved) from infusion (figure 2B,C). No patients developed secondary malignancies following CD19.22.BBζ.

All 28 patients had evidence of PB CAR T-cell expansion, peaking at a median of 7 days (range, 6–15) with a median peak level of 48.2 CAR T cells/mcL (range, 3.4–1,614 CAR T cells/mcL) (online supplemental figure 1A). Expansion was higher in CRS+ patients (median 75 CAR T cells/mcL vs 15.9 CAR T cells/mcL, p=0.016) and with high-disease burden (≥M2 marrow) (median 105.9 CAR T cells/mcL vs 21.1 CAR T cells/mcL, p=0.0002), with no differences based on non-CNS EMD (online supplemental figure 1B-D). CAR T cells remained detectable in 16 (57.1%) patients at day 28 in the PB, with a median of 5.5 CAR T cells/mcL (range, 1.7–205.3) (online supplemental figure 1E). B-cell aplasia was seen in all 25 responders at day 28.

At BM restaging, CAR T cells were detected in 19/27 (70.4%) patients at a median of 0.82% of T cells positive for CAR (range, 0–67%) (online supplemental figure 1F). Across 27 patients with a post-infusion lumbar puncture, 18 CSF samples were sufficient to evaluate for CAR T cells and 14 (77.8%) had detectable CAR. The three patients with grade 3 ICANS had early CSF analysis on day 5–7 demonstrating high levels of CAR T cells (median 57.3%) with persistently detectable CAR T cells in the CSF at day+28 despite intrathecal and systemic corticosteroids (online supplemental figure 1G). Similarly, for the one patient who received siltuximab, CAR T-cell expansion and persistence remained comparable to the rest of the cohort (online supplemental figure 2).

Cytokine profiles demonstrated increased innate and interferon (IFN)-γ associated biomarkers (eg, IFN-γ, IL-6, IL-10, chemokine (C-X-C motif) ligand (CXCL10)) in CRS+patients compared with those without CRS (figure 2D). Elevations occurred mainly in patients requiring tocilizumab and returned to levels similar to patients without CRS following therapeutic intervention and/or time (online supplemental figure 3). Interestingly, IL-18 levels continued to increase after CRS-directed therapy and trended upwards in all patients with CRS even after CRS resolution. Additionally, in one patient that received siltuximab for CRS, cytokine responses post-treatment followed a different pattern than individuals treated with tocilizumab, including a spurious rise in IL-6, a finding previously reported following siltuximab in multicentric Castleman disease (online supplemental figure 4).

All infusion products at the RP2D were successfully manufactured. The median transduction efficiency was 68.0% (range, 46.8–87.5%) based on Protein L staining at harvest, with median fold expansion at harvest of 14.6 (range, 6.8–20.9). While CD4:CD8 ratios in the leukapheresis product were 0.98 (range, 0.3–7.3), the infusion product had CD4 predominance, with a CD4:CD8 ratio of 2.0 (range, 0.4–5.1), consistent with similarly manufactured CAR T-cell products.

Additional flow cytometry analyses on a subset of available infusion products (n=12; all of whom achieved CR) showed that products were composed of 37.8% T memory stem cells, 9.6% central memory T cells, 16.6% effector memory T cells, and 36.2% effector T cells (online supplemental figure 5A). Although gene expression analysis showed that products clustered by baseline disease burden, with overexpression of genes associated with T-cell activation and the Th17 axis observed in the high-disease burden group (online supplemental figure 5B and C), transduction efficiency, vector copy number, CD4:CD8 ratios, and T-cell phenotypes were similar between those with high and low-disease burden. Results from T-cell phenotype characterization, metabolic analyses, and gene expression analyses are detailed in the online supplemental Appendix.

The bivalent CD19.22.BBζ CAR T-cell represents our institution’s first venture into dual antigen targeting. The construct was highly effective and generally well-tolerated with an MRD-negative CR rate of 89.3% at the RP2D. This is aligned with response rates seen with single antigen targeting of CD19 and/or CD22 CAR T cells. Failure to achieve remission occurred only in patients with baseline non-CNS EMD, and even in this population, 57.1% had full disease eradication. This experience adds to data suggesting that, relative to BM and CNS disease, there may be limitations to CAR T-cell efficacy in non-CNS EMD in B-ALL, including with single-antigen targeting. Whether poor CAR T-cell trafficking or the immunosuppressive microenvironment of specific tissues prevent optimal eradication in non-CNS EMD warrants further study. Beyond this, CD19.22.BBζ proved an effective bridge to HSCT, with 82.1% going on to consolidative HSCT, and 16 (57.1%) alive nearly 2 years post infusion. Despite limitations of CD22 targeting, likely due to its large extracellular domain and potential spatial steric hindrance, CD22 downregulation occurred in at least one patient at the time of relapse. Direct comparison of antigen escape patterns with historical single-target cohorts is limited by lack of longitudinal antigen density quantification.

The safety and tolerability of CD19.22.BBζ is notable, with predominantly low-grade CRS, with six responders having no evidence of CRS, illustrating that CAR T cells can be efficacious without clinically significant systemic inflammation. Cytokine profiling revealed distinct patterns associated with CRS incidence, including elevations in IL-18 at 2 weeks post-infusion in patients who developed CRS, similar to the pattern observed with our CD22 CAR trial. Nonetheless, the present patients did not present with any clinical or laboratory features of IEC-HS, warranting further study into the role of IL-18. Grade 3 ICANS occurred in three patients with high-disease burden, but was rapidly reversible.

Following three cases of ICANS (two of whom received antecedent tocilizumab), the study was amended to test the safety and efficacy of siltuximab as a first-line agent in the treatment of CRS. Anecdotal evidence has suggested that ICANS may be potentiated or worsened by tocilizumab due to the subsequent rise in serum and CSF IL-6 following inhibition of the IL-6 receptor. Indeed, our CD19-28ζ trial demonstrated that patients who received tocilizumab had elevated levels of CSF IL-6 on day+28 compared with those that did not. Although we were only able to test siltuximab in one patient with high-disease burden and grade 3 CRS, two doses fully resolved CRS without subsequent ICANS or impact on CAR T-cell expansion or persistence. Our data, alongside preliminary adult data suggesting a role for siltuximab in rapid CRS resolution, supports further study.

Given the favorable safety profile and efficacy of CD19.22.BBζ, this trial remained open until activation of our subsequent bicistronic CD19.28ζ/CD22.BBζ construct, which is currently enrolling in a phase 1/2 trial (NCT05442515). The design of this next-generation construct directly addresses the limitations observed with the CD19.22.BBζ construct, particularly suboptimal CD22 engagement. Preliminary clinical experience suggests effective dual targeting of CD19 and CD22 with substantial antitumor efficacy, including in those with non-CNS EMD. Enrollment is ongoing.