CD19-CAR T cells persist long-term in the cerebrospinal fluid of pediatric, adolescent and young adult patients with B-cell acute lymphoblastic leukemia

The central nervous system (CNS) is a sanctuary site for B-cell acute lymphoblastic leukemia (B-ALL), particularly in the relapsed/refractory (r/r) setting [ 1 ]. Historical treatments center on high-dose systemic/intrathecal chemotherapy and/or focal irradiation, which come with high-risk of long-term side effects [ 2 , 3 , 4 ]. CD19-CAR T-cell therapy has become an invaluable therapeutic option for pediatric, adolescent and young adult (AYA) patients with r/r B-ALL, including for those with CNS disease [ 5 , 6 , 7 , 8 , 9 ]. CD19-CAR T cells can persist long-term in the peripheral blood (PB) and contribute to maintenance of remission. PB persistence is commonly evaluated using either functional readouts of ongoing autologous B-cell aplasia or by quantitative PCR (qPCR) assays. However, it is currently unknown how long CD19-CAR T cells persist in the CNS. To address this question, we longitudinally analyzed the cerebrospinal fluid (CSF) of 29 pediatric and AYA patients for the presence of CD19-CAR T cells, using a previously developed qPCR assay [ 10 ]. All patients received lymphodepleting chemotherapy (fludarabine, cyclophosphamide) and tisagenlecleucel (tisa-cel) for relapsed/refractory B-ALL between 10/2020 and 1/2025, as standard of care, and were enrolled on a St Jude Children’s Research Hospital institutional review board (IRB)-approved correlative study protocol. Written informed consent/assent for treatment and participation in this study were obtained from patients/guardians in accordance with institutional guidelines and the Declaration of Helsinki. CNS status within 4 weeks prior to CAR T-cell infusion was determined via lumbar puncture (LP), with historical maximum CNS status recorded. Post infusion cytokine release syndrome (CRS) and immune effector cell associated neurotoxicity syndrome (ICANS) were graded using the ASTCT grading system [ 11 ]. Disease response was evaluated approximately 4 weeks post-CAR T-cell infusion, with minimal residual disease (MRD)-negative (neg) defined as no detectable disease in the bone marrow (BM) by flow cytometry, PCR, and/or next generation sequencing (NGS; ≤10 copies/million). Patients remained on study for sample collection until time of initial non-response, development of recurrent detectable disease, death, receipt of a consolidative allogeneic hematopoietic cell transplant (HCT), or completion of the 12-month study endpoint, whichever occurred first (data cutoff, 9/10/2025). Figure generation and statistical analysis were performed using GraphPad PRISM (v10.6). Twenty-nine patients are included, with patient and disease characteristics summarized in Table 1 . The median age at time of cellular infusion was 7 years (range, 1–20); most patients identified as male ( n = 19). Prior to CAR T-cell infusion, no patients had detectable CNS disease; 20 had a history of CNS B-ALL. Prior leukemia-directed therapy for this patient cohort was diverse, including 7 patients with prior CNS-directed radiation therapy, 3 of which occurred within 3 months prior to CAR T-cell therapy. Systemic and CNS-directed bridging therapy was determined by the treating physician, with most patients receiving intrathecal chemotherapy with hydrocortisone and cytarabine within 2 weeks prior to the start of lymphodepleting chemotherapy. Post infusion, 21 patients developed CRS (grade 3–4, n = 3) and 10 developed ICANS (grade 3–4, n = 5). Twenty-seven of 29 patients achieved disease response in the BM post infusion (CR-MRDneg, n = 22), with no patients having detectable CNS disease. Post response assessment, 8 patients proceeded to consolidative HCT (median 56.5 days, range 43–84), 7 had recurrent disease (median 84 days, range 49–804) and 1 developed treatment-related myelodysplastic syndrome (MDS). Amongst the 7 with recurrent disease, 4 were CD19-negative, 1 CD19-positive, and 2 had unknown antigen status (detectable by NGS only). Among those with CNS involvement at time of disease recurrence ( n = 3), 1 patient experienced an isolated CNS relapse (CD19-neg) and 2 had combined BM/CNS relapse (CD19-neg, n = 1). Of note, 2 of these patients had a history of CNS involvement. Table 1 Patient, disease and treatment characteristics. Full size table qPCR analyses for the CD19-CAR transgene and the T-cell receptor α constant (TRAC) locus, were performed on PB, BM, and CSF samples at 1-, 2-, 3-, 6-, 9- and 12-months post infusion. We detected CD19-CAR T cells in 28 of 29 CSF samples (96.4%) at 1 month post tisa-cel infusion, and in 41 of 49 samples (80.5%) at later time points (Fig. 1A, B ). Most CSF samples (58.2%) contained no red blood cells (RBCs) (Fig. 1C ). Importantly, presence of CD19-CAR T cells did not correlate with the presence of RBCs, suggesting that the detected CD19-CAR T cells were CNS resident (Fig. 1D ). Similarly, the ratio of CD19‑CAR transgene to TRAC locus copy number per μg of genomic DNA was significantly higher in the CSF than in PB ( p < 0.0001), indicating that CD19-CAR T cells are enriched in the CSF on a per cell basis, providing additional evidence that our qPCR assay detected CSF-resident CD19-CAR T cells (Fig. 1E ). While CD19-CAR T-cell levels correlated in the PB and BM ( R 2 = 0.5645) (Fig. 1F ), no relationship between the PB and CSF was observed ( R 2 = 0.2633) (Fig. 1G ), highlighting the role of the blood-brain-barrier (BBB) as a gate keeper. Finally, patients with and without a history of CNS disease had similar levels of CAR T-cell detection in the CSF at 1-month post infusion (median copies/µg DNA: CNS1, 1197; CNS2: 599; CNS3: 796). Fig. 1: Long-term persistence of CD19-CAR T cells and pleocytosis in the cerebrospinal fluid (CSF) post CD19-CAR T-cell therapy. Full size image A , B Detection of CD19-CAR T cells 1 to 12 months post CAR T-cell infusion. C Red blood cell (RBC) contamination in CSF samples. D RBCs vs CAR T cells. E CAR:TRAC ratio in peripheral blood (PB) vs CSF. Correlation of CAR T cells in PB vs ( F ) Bone Marrow (BM) or ( G ) CSF. H Total nucleated cells (TNC) in the CSF vs RBCs. I TNC CSF differential vs RBCs. J TNC in CSF 1 to 12 months post CAR T-cell infusion. K CAR T cells vs TNC in CSF. L CAR T cells vs <3 or ≥3 TNC in CSF. M CAR T cells in the CSF vs CRS grade. N CAR T cells in the CSF vs ICANS grade. O CAR T cells in PB vs CRS or ICANS grade. Samples from all patients are shown ( n = 29); qPCR data is presented as copies per μg of genomic DNA; B , D , K – O : *negative qPCR results were assigned a value of 10 for plotting purposes; L – O : shaded area: lower limit of detection (LOD) of qPCR assay; non-parametric test (Mann-Whitney test) except for E : paired t-test, J : two-way ANOVA, and F , G : linear regression analysis; qPCR data was log10 transformed before analysis; ns: not significant, * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. Twenty patients had long-term serial CSF samples available for analysis. Of these, 7 patients had no detectable CAR T cells at last CSF evaluation and 1 subsequently experienced isolated CNS relapse (CD19-neg). Of note, all 7 of these patients concurrently had detectable CAR T cells in the PB. Among the 13 patients with detectable CAR T cells in the CSF at last analysis, two subsequently experienced CNS relapse. However, as these relapses occurred 1 month and >1 year after the last CSF assessment, it is unknown whether CAR T cells were still present at the time of relapse. As reported by others [ 12 ], pleocytosis in the CSF was dominated by lymphocytes and independent of RBC contamination (Fig. 1H–J ). While there was no direct correlation between increasing CSF total nucleated cell (TNC) count and number of CD19-CAR T cells in the CSF (Fig. 1K ), samples with a CSF TNC count of ≥3/μL did have a significantly higher number of CD19-CAR T than CSF samples with a TNC count of <3 cells/μL ( p = 0.0008) (Fig. 1L ). Interestingly, patients with long-term persistence of CAR T cells in the CSF also had long-term persistent CSF pleocytosis. Pleocytosis has also been reported in patients receiving Blinatumomab [ 13 ]. Clearly more studies are needed to understand the biology and role of pleocytosis in patients receiving immunotherapies. We next investigated if the history of CRS or ICANS influenced frequency of CD19-CAR T cells within the CSF at 1 month post infusion. Patients with higher grade (2–4) CRS ( n = 10) or ICANS ( n = 6) had a significantly greater level of detection of CD19-CAR T cells by QPCR in the CSF than patients with lower grade (0–1) CRS ( p = 0.0001) or ICANS ( p = 0.0412) (Fig. 1M, N ). Similar analysis for PB samples only revealed a significant difference for patients with higher grade CRS ( p = 0.0122) (Fig. 1O ). This suggests that systemic inflammation can influence CD19-CAR T-cell expansion and migration into CSF. Among the 6 patients with grade 2–4 ICANS, 4 had a history of CNS leukemia; this finding was similar to that seen with grade 0–1 ICANS, where 16 of 23 patients had a history of CNS leukemia. Finally, we analyzed cytokines, chemokines, and growth factors using a 38- or 47- Multiplex panel in the CSF at 1 month post infusion ( n = 28) (Figs. S1 , S2A ). Mean cytokine and chemokine concentrations were less than 10 pg/mL except for Fractalkine, IP-10, MCP-1, MIG, IL8, and IL27. Combining all cytokines and chemokines into a single value (CytoChemo Sum analysis) revealed no correlation with CSF pleocytosis, the number of CAR T cells, or a history of higher-grade CRS or ICANS (Fig. S2B–E ). To our knowledge, we report for the first time longitudinal data on the persistence of CD19-CAR T cells in the CSF of pediatric and AYA patients post tisa-cel infusion. Our findings highlight a potential advantage of CD19-CAR T cells over other immunotherapy approaches, such as Blinatumomab, which has limited CNS penetration, a short half-life and higher rates of extramedullary relapse post therapy, including in the CNS [ 4 , 14 , 15 ]. However, due to the limited number of CNS relapse events after CD19-CAR T-cell therapy in our cohort, we were unable to correlate this outcome with CSF CAR T-cell persistence. Future studies are needed to confirm our findings in a larger cohort of patients, gain mechanistic insights, and explore potential surrogates for CAR T-cell persistence that are more widely available than qPCR.