Chimeric antigen receptor (CAR)-T cell therapy has yielded remarkable clinical efficacy in relapsed or refractory (r/r) leukemia (1, 2). Our previous phase II clinical study confirmed the safety and efficacy of CD19/22 co-administration (3). However, cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS) occur with incidence rates of 80%–90% and 20%–72%, respectively, restricting the further promotion of CAR-T cell therapy as a viable treatment option (3–5). CRS management primarily involves anti-cytokine therapy to prevent disease progression or even life-threatening conditions (6–8). The interleukin-6 (IL-6) receptor antagonist tocilizumab is an important treatment for patients with CRS, but it does not cross the blood–brain barrier and tocilizumab may precipitate ICANS in high-risk patients due to the transient increase in the systemic and central nervous system (CNS) IL-6 levels following tocilizumab administration (9, 10). Glucocorticoids (GCs) have a powerful ability to suppress the immune response and are effective in preventing the development of severe CRS or ICANS. GCs have been used in 53% of patients with CRS (11). However, their use can severely impair T-cell function and proliferation (12) and can inhibit CAR-T cell function and induce apoptosis, potentially impairing the effectiveness of CAR-T cell therapy; thus, guidelines do not recommend GCs as a first-line treatment (6, 12, 13). The use of corticosteroids for toxicity management in CAR-T therapy has raised concerns about whether such treatment may affect the anti-leukemic effects of CAR-T cell therapy. Therefore, we conducted a retrospective study to explore whether the use of GCs affects the efficacy of CAR-T cell therapy against leukemia, which will provide a clinical reference for therapy for CAR-T-cell-related toxicities and will contribute to the wider application of CAR-T cell therapy in the treatment of leukemia.

This retrospective study included 193 patients aged ≤18 years and 1 younger adult aged 19.6 years who had r/r B-cell acute lymphoblastic leukemia (B-ALL) who were treated at our center with CD19/22 CAR-T cells between September 2019 and December 2021 (Trial Registration No. ChiCTR2000032211, www.chictr.org.cn), as shown in Supplementary Figure S1. We reviewed the medical records of all the enrolled cases and collected information, including grade, duration, and dose of corticosteroids used, and outcome. CRS and ICANS were prospectively evaluated in accordance with the American Society for Transplantation and Cellular Therapy (ASTCT) consensus grading related to immune effector cells, as detailed in Supplementary Tables S1, S2 (13). Medication for CRS and ICANS was administered in accordance with toxicity guidelines for pediatric patients (8).

The treatment of toxicity in CAR-T therapy is an emerging topic, with ongoing research aimed at enhancing safety while preserving the longevity of the therapeutic benefit. This retrospective investigation posits that steroids do not affect the remission rate of CAR-T cell treatment in leukemia, consistent with the findings of Liu et al. (15). Low-dose corticosteroids were employed as first-line treatment to manage CRS in patients with B-ALL undergoing CD19- or CD22-CAR-T cell therapy, encompassing both children and adults; however, a comparison of survival between the two groups was not conducted.

This study, by categorizing varying disease burdens as seen in Figures 1B, 2B, indicates that corticosteroid administration did not influence the OS of patients undergoing CAR-T cell treatment for leukemia. This contrasts with the findings of Strati et al. (16), who retrospectively analyzed steroid usage in 100 adults with large B-cell lymphoma (LBCL) undergoing CAR-T cell therapy, with 60 subjects utilizing corticosteroids while 40 did not, and concluded that steroid use significantly diminished overall survival. The discrepancies in results may be attributed to the following factors. First, an examination of the research participants reveals that our study focuses on pediatric patients with B-ALL, all under the age of 20, whereas the international research subjects consisted of adults with LBCL. Research indicates that the process by which CAR-T cells eliminate tumor cells is primarily governed by three axes: the perforin and granzyme axis, cytokine secretion, and the Fas and FasL axis (17, 18). Larson et al. (19) discovered that solid tumors and liquid tumors exhibit distinct interaction mechanisms with CAR-T cells, indicating that solid tumors with defects in the IFN-γ receptor gene demonstrate greater resistance to CAR-T cell cytotoxicity, whereas liquid tumors are not rendered insensitive to CAR-T cells due to IFN-γ receptor gene defects. One mechanism by which GCs suppress the immune response is through the inhibition of T-cell activity by reducing cytokine production, including IFN-γ (12). We hypothesize that this may be attributed to the differing interactions of CAR-T cells with solid tumors vs. leukemia; CAR-T cells that target solid tumors depend on the IFN-γ receptor signaling pathway for cytotoxicity, whereas leukemia cells exhibit reduced reliance on this pathway. The use of modest dosages of corticosteroids did not impact the overall survival of leukemia patients; nevertheless, this conclusion requires to be validated via further studies.

Conversely, our center adheres to stringent standards regarding the indication for corticosteroid use, as outlined in the Methods section. Consequently, only 44% of patients with CRS utilized corticosteroids, with a median cumulative dexamethasone-equivalent dosage of 0.56 mg/kg; the majority were administered a single dose, and corticosteroids were promptly withdrawn upon improvement of the clinical symptoms. Only three individuals experienced persistent grade 4 CRS due to the ongoing use of high-dose methylprednisolone.

An in vitro study by Brummer et al. (20) that treated glioblastoma cell lines with CAR-T cells and dexamethasone demonstrated the efficacy of high concentrations of dexamethasone to antagonize CAR-T cells by depleting or reducing the activity of CAR-T cells and promoting tumor cell growth. The cumulative dosage of dexamethasone exceeding a threshold level may influence CAR-T cell functionality. In a mouse validation experiment (21), researchers discovered that high dosages of dexamethasone (>5 mg/kg) abolished CAR-T-cell-mediated tumor eradication, but CAR-T cell tumor lysis was sustained at lower doses of 0.2 or 1 mg/kg of dexamethasone. Our investigation indicates that, as seen in Figures 1E,F, elevated dosages of dexamethasone (more than 0.56 mg/kg) impaired EFS and OS in the LDB group; nevertheless, the difference lacks statistical significance. Furthermore, only three individuals received a cumulative dose of dexamethasone greater than 5 mg/kg. Consequently, we hypothesize that the administration of dexamethasone in the patients in this trial may not have attained the dosage required to influence CAR-T cell impairment.

The timing of corticosteroid administration in the treatment of clinical toxicity remains a contentious topic. Strati et al. (16) posited that corticosteroid administration should be postponed as long as the CRS response is manageable. Some research studies suggest that early corticosteroid intervention may mitigate the risk of severe CRS and ICANS, benefiting patients without compromising therapeutic effectiveness (15, 22, 23). We observed that individuals who were administered corticosteroids exhibited more favorable outcomes during severe grade 3 or 4 CRS in the HDB group (Supplementary Figure S4C). The HDB group also included nine patients who did not receive corticosteroids and the biological and clinical characteristics of this group are detailed in Supplementary Table S3. Three patients had a risk gene (MLL rearrangement positive). We propose that this may be a contributing factor to the worse prognosis.

Disease recurrence after remission with CAR-T cell therapy remains challenging, with several studies indicating that the depletion of CAR-T cells is a primary factor contributing to tumor recurrence (24). The use of corticosteroids following CAR-T cell infusion in certain individuals did not influence the growth or durability of CAR-T cells (Figure 3). The quantity of CAR-T cells in the steroid group was marginally higher than in the non-steroid group, potentially attributable to their elevated tumor burden. The increased antigen stimulation from CAR-T cell infusion may lead to a significant increase in CAR-T cells, resulting in a more pronounced CRS response, thereby increasing the likelihood of corticosteroid administration.

Greater tumor burden was linked to an elevated likelihood of severe CRS (25). In our LDB group, those who did not receive corticosteroids or were administered lower doses (Figures 1C,E) exhibited superior EFS. However, Oluwole et al. (26) administered dexamethasone prophylactically to LBCL patients on the day prior to CAR-T cell infusion and they discovered that preventive and early corticosteroids treatment resulted in no grade 3 or higher CRS but had a significant response rate. Consequently, corticosteroid utilization can be reduced when the disease burden is low; conversely, a high disease burden increases the risk of grade 3 or 4 CRS reactions. Early administration of corticosteroids may facilitate prompt management of life-threatening CRS reactions, enabling immediate cessation of the treatment once the reaction is clinically manageable.

These data suggest that the judicious application of corticosteroids, in accordance with existing CRS prevention and control protocols, may not influence the clinical effectiveness or overall survival of r/r B-ALL patients undergoing CAR-T treatment. In CAR-T cell therapy for leukemia patients, a clinician should avoid administering corticosteroids when the disease burden is low. Conversely, when the disease burden is high, the risk of grade 3 or 4 CRS reactions increases, and corticosteroids may be administered early to prevent life-threatening CRS reactions until the clinical symptoms improve, after which they should be discontinued immediately.