Abstract Introduction CD7-targeted chimeric antigen receptor (CAR) T-cell therapy has demonstrated efficacy and safety in treating T-cell acute lymphoblastic leukemia/lymphoma (T-ALL/LBL). However, evidence remains limited regarding the ability of CD7 CAR-T cells to cross the blood-brain barrier and effectively eradicate central nervous system (CNS) involvement in T-ALL/LBL without exacerbating neurotoxicity. Here we evaluated the efficacy and safety of CD7 CAR-T cells in pediatric and adult patients with relapsed/refractory (R/R) T-ALL/LBL involving the CNS through phase I/II clinical trials (NCT04572308 and NCT04916860). Methods Peripheral blood mononuclear cells were collected via leukapheresis from enrolled patients. A novel approach was used to generate “naturally selected” CD7 CAR (NS7CAR)-T cells that overcome CD7-directed fratricide without gene editing. NS7CAR is a 2nd-generation murine-based CAR-T containing 4-1BB and CD3ζ co-stimulatory domains. Patients who relapsed after allogeneic hematopoietic stem cell transplantation (allo-HSCT) without active graft-versus-host disease (GVHD) were eligible. All patients received intravenous fludarabine (30mg/m2/d) and cyclophosphamide (300mg/m2/d) (FC) conditioning regimens on day -5 to day -3 prior to NS7 CAR-T infusion. To manage rapidly progressive disease, bridging therapy was allowed between leukapheresis and FC conditioning. CD7 CAR-T cell proliferation and blast clearance in cerebrospinal fluid (CSF) were monitored using flow cytometry (FCM). Results Thirty patients with CNS involvement (24 T-ALL and 6 T-LBL) received NS7CAR T cell infusion. The median age was 23 years (range, 4-44). At enrollment, 13 patients were classified as CNS-3 (WBC ≥5/mcL in CSF), and 17 as CNS-2 (WBC 5/mcL in CSF). Seven patients relapsed from prior allo-HSCT. For high-risk features, 7 patients harbored STIL::TAL1 fusion genes, and 1 patient had TP53 gene mutations. Nineteen patients had blasts in bone marrow (BM), with a median blast of 38%(range:0.5-93%), including 4 patients with blasts80%. Non-CNS extramedullary disease (EMD) was present in 10 patients. Bridging therapy was administered in 25 patients. All patients received a single dose of NS7 CAR-T cells at a low dose (5×105cells/kg, n=2), and a medium dose (1-1.5×106cells/kg, n=28). Following infusion, 28 patients had available CAR-T cell expansion data in CSF. The median maximum level of NS7CAR-T cells in CSF was 40.5% (range, 0.0-94.9%), which occurred on day 19 (range, 13-51) as detected by FCM. About 1 month post CD7 CAR-T infusion (except 1 evaluated on day 51), 30/30 (100%) patients achieved complete remission (CR) in CSF, and 17/19(89.5%) patients showed minimal residual disease (MRD)-negative CR in BM. For non-CNS EMD, the overall response rate was 9/10 (90%, 6 CR and 3 partial remission). The median follow-up time was 276.5 days (range, 58-1676). The 1-year overall survival (OS) and progression-free survival (PFS) was 83.0% and 52.8%, respectively, and the 3-year OS and PFS were 67.1% and 45.3%. Among the 24 patients achieving CR in both EMD and BM, 18 received consolidation allo-HSCT within 3 months post-CAR-T, with the 1-year OS and PFS of 87.%% and 75.1% and the 3-year rates of 70.0% and 60.1%. Of the remaining 6 CR patients, 4 relapsed within 3 months, 1 was lost to follow-up on day 180, and 1 died on day 75. Eight patients relapsed following CAR-T infusion (6 in BM and 2 in CNS) at a median of 65 days (range, 33-265), of whom 5 patients with STIL::TAL1 fusion genes relapsed within 3 months. Three of 8 patients lost CD7 expression. Specifically, both CNS relapsed patients experienced early relapse on day 56 and day 72, respectively. The majority of (27/30, 90%) patients experienced mild cytokine release syndrome (CRS, grade I, n=26 and grade II, n=1), and 3/30(10%) patients had grade III CRS. Neurotoxicity occurred in 2 patients (6.7%, grade IV), and others had no neurotoxicity. All were effectively managed with corticosteroids and/or tocilizumab. Conclusion This study establishes NS7CAR-T cells as a promising therapeutic option for R/R T-ALL/LBL with CNS involvement, demonstrating a manageable safety profile. Notably, disease relapse primarily occurred at non-CNS sites. We identified STIL::TAL1 fusion as a potential predictor of early relapse. Further studies with extended follow-up and larger cohorts are warranted to better define the long-term efficacy and durability of NS7CAR-T therapy in CNS-involved T-ALL/LBL.
Zhang et al. (Mon,) studied this question.