Real-World Assessment of Prophylactic Anakinra on Neurotoxicity after CD19 CAR T-cell Therapy in R/R B-cell Lymphoma: An Inverse Probability of Treatment Weighting Analysis

Immune effector cell-associated neurotoxicity syndrome (ICANS) is a common toxicity of CD19 CAR T-cell therapy (CAR T). Anakinra, an IL-1 receptor antagonist, has been used prophylactically to prevent ICANS, but real-world data are limited. Our center implemented anakinra prophylaxis for high-risk patients (age ≥65 or CD28-based CAR T). We conducted an inverse probability of treatment weighting (IPTW) analysis comparing R/R B-cell lymphoma (R/R BCL) patients who received anakinra prophylaxis to historical controls. Adults with R/R BCL receiving second line or greater CAR T from 2018-2025 were included. Controls were those who would have received anakinra under current policy. Anakinra 100 mg q12h was started day 0, titrated per physician, and continued until discharge. The primary endpoint was grade ≥3 ICANS. Secondary endpoints were CRS (any, grade 3+), ICANS (any), CRS/ICANS duration, infection (any, grade 3+), and survival. IPTW used a propensity model including age, disease, LDH, prior transplant, ≥3 lines, KPS ≤70, construct, CNS/extranodal disease, and disease status. Balance was confirmed by standardized mean differences. Weighted logistic regression estimated odds ratios (ORs) for outcomes, controlling for the same variables. Among 178 patients (67 anakinra, 111 control), covariates were balanced post IPTW (Table 1). Grade ≥3 ICANS rates were similar between anakinra and controls (12.2% vs. 9.6%; p = 0.62), but anakinra patients had more all grade ICANS (40.5% vs 22.8%, p = 0.03) and grade ≥3 CRS (11.9% vs 3.4%; p < 0.05) (Figure 1), with similar rates of any grade CRS (p = 0.74). CRS duration (3 vs 4 days; p = 0.47) and ICANS duration (4 vs 5, p = 0.81) were similar. Anakinra patients got more dexamethasone for CRS or ICANS (56.1% vs 28.8%; p < 0.001) and tocilizumab (77.3% vs 58.1%; p = 0.02) Among dexamethasone recipients, there was no difference in total dose (100 vs 180 mg; p = 0.40). Grade ≥3 infections by day 30 were higher with anakinra (20.8% vs 7.8%; p = 0.03), driven by grade ≥3 bacterial infections (16.1% vs 7.8%; p = 0.14) and bacteremia (13.4% vs 4.2%; p < 0.05). In IPTW regression, anakinra was not associated with grade ≥3 ICANS (p = 0.28) but increased odds of any grade ICANS (OR 3.3; 95% CI 1.4-7.6; p = 0.006), grade ≥3 CRS (OR 18.2; 95% CI 1.3-254.9; p = 0.03), any grade infection (OR 3.5; 95% CI 1.3-8.9; p = 0.01), and grade ≥3 infection (OR 7.4; 95% CI 1.9-28.7; p = 0.004). Response rates (ORR 87% vs 92%, p = 0.41; CRR 72% vs 82%, p = 0.30) and survival (EFS p = 0.28; OS p = 0.72) were similar. Multivariable Cox models showed no survival impact from anakinra (EFS p = 0.19; OS p = 0.76). Prophylactic anakinra did not reduce severe ICANS but was associated with higher rates of any grade ICANS/grade ≥3 CRS and severe infections. These findings caution against routine anakinra prophylaxis and suggest the need for further prospective evaluation.