Safety and Feasibility of Intrathecal Infusion of CD19 and/or CD22 CAR T Cells in r/r B-ALL Patients with CNSL or High-Risk CNS Relapse

The central nervous system (CNS) is a sanctuary for leukemia and approximately 10% of patients have pre-existing CNS infiltration at diagnosis. Conventional central nervous system leukemia (CNSL) therapy included high-dose chemotherapy, intrathecal drugs and cranial radiotherapy. However, challenges remain in CNS relapse after multiple chemotherapy and stem cell transplantation. A novel treatment for relapsed or refractory B-cell acute lymphoblastic leukemia (r/r B-ALL) using CD19 or CD22 chimeric antigen receptor (CAR) T cells has recently shown promising results. Considering CAR T infusion may cause severe immune effector cell-associated neurotoxicity syndrome (ICANS) if the CNS tumour burden is too high, most of available clinical trials excluded CNSL. One historical comparison consist of seven r/r B-ALL patients selected fromNCT04340154 with baseline CNS3 achieved complete remission (CR) both in bone marrow (BM) and CNS after sequential CD19 and CD22 CAR T cell. Six and one patients had grade 1cytokine release syndrome (CRS) and ICANS, respectively. Two patients (29%) had CNS relapse, including one patient who relapsed at 5.4 months with undetectable CAR in cerebrospinal fluid (CSF), while the other relapsed at 8 months with disappearing CAR in CSF and peripheral blood (PB). Based on this, we assumed patients who had baseline CNSL had high risk CNS relapse after CAR which might be related to disappear of CAR in CSF. Intrathecal infusion of CAR T might prolong the CAR persistence in CSF and CSF remission. Therefore, We performed a retrospective analysis as a pilot study to determine the toxicity and effectiveness of intrathecal CD19 and/or CD22 CAR T in 26 r/r B-ALL patients with CNSL or CNSL history after inducing CR/CRi or lower CSF tumour burden by prior intravenous CAR T. From November 2021 to May 2024, 14, four and eight patients who were classified as CNS1, CNS2 and CNS3 before lymphodepletion received intravenous plus intrathecal CAR T infusion. Twenty received intrathecal infusion of CD19 and 18 received CD22 CAR T cells at a median time of 15 days (range 14-101) and at 27 days (range 15-46) after intravenous CD19 or CD22 CAR T , respectively. Grade 1-2 CRS, ICANS, and grade 3 ICANS occurred in 18 (69%) , five (19%), and one (4%) patients after intravenous CAR T infusion and in eight (31%), four (15%) and one (4%) patients after intrathecal CAR T infusion, respectively. All intravenous infusion related CRS and ICANS recovered before intrathecal infusion. Of 14 patients without baseline CNSL , five (36%), two (14%) and one (7%) had grade 1 CRS, ICANS and grade 3 ICANS after inthrathecal CAR T infusion. Of 12 patients with baseline CNS2-3 status, three (25%), and two (17%) had grade 1 CRS and ICANS after inthrathecal CAR T infusion. The median onset time of CRS after inthrathecal CAR T infusion was day 2 (range 1-3) with median persistence time of 1 day (range 1-3) and ICANS was day 1(range 1-1) with 2 days (range 1-4). The common CNS symptoms were headache (4, 15%), dizziness (3, 12%), somnolence (1, 4%), tremor (1, 4%) and delirium (1, 4%). Most CRS and ICANS were spontaneous relieved except one patient with grade 3 ICANS on days 10-13 after intrathecal CD19 CAR T characterized by headache, dizziness, nystagmus and nuchal rigidity. 21 of 22 patients with active BM and/or CNS disease at baseline achieved BM and CNS CR/CRi at a median time of 15 days (range 14-51) after initial intravenous CAR T cells. One patient with CNS3 plus BM disease had no response to both intravenous and inthrathecal CAR T cells. Four patients with non-active disease maintained CR both in BM and CNS. With a median follow up of 10 months (range 1.8-28), 20 patients were in continuous CR. Of 14 patient with CNS1 status at baseline, two had BM relapse at 5.2, 13.5 months and one had BM plus CNS relapse at 7 months. Of 12 patients with CNS 2-3 at baseline, one had CNS relapse at 8.9 months and the other had BM relapse at 5.4 months. This analysis in 26 r/r B-ALL with CNSL or high-risk CNSL relapse demonstrated the safety and feasibility of intrathecal CAR T. The incidence of CRS and ICANS were comparable between intravenous and intrathecal CAR T. The incidence of CNS relapse in patients with baseline CNS 2-3 in this analysis (8%) was lower than that in NCT04340154 (29%). However, the CAR persistence in CSF was not clarified due to limited available data. Intrathecal after intravenous CAR T infusion may be a feasible method for CNSL, but need to be verified in a larger sample size.