Central nervous system relapse is uncommon after chimeric antigen receptor T-cell therapy for relapsed/refractory large B-cell lymphoma

Introduction: Chimeric antigen receptor (CAR) T-cell therapy has significantly improved the prognosis of patients with relapsed/refractory (R/R) large B-cell lymphoma (LBCL). However, over 50% of patients will experience progressive disease (PD) after infusion. While relapse in the central nervous system (CNS) is uncommon (<5%) after chemoimmunotherapy (CIT), large real-world cohorts focused on this specific event after CAR T-cell therapy are lacking. Here, we carried out a comprehensive evaluation of the CNS relapse risk in the post-CART space and compared the results to a CIT cohort. Methods: We conducted a multicenter, international study including patients with R/R LBCL treated at 9 centers with commercially-available CAR T-cell products until June 2025. For the comparative cohort, we included transplant-eligible patients who received second-line CIT in the GELTAMO network until December 2023; if these patients received CAR T-cell therapy at a later line, they were censored. Patients with a prior history of CNS involvement were excluded from the study. The primary endpoint was incidence of CNS relapse in the CAR-T cohort. Secondary endpoints included CNS relapse among patients with high-risk CNS-IPI, with 1 prior line and in those who experienced PD after CAR-T; the same endpoints were assessed in the CIT cohort. Results: This study included 1102 R/R LBCL patients; 875 (79%) and 227 (21%) comprised the CAR-T and CIT cohorts, respectively. Median follow-up for the full patient population was 41 months (95%CI 37-46). Focusing on the CAR-T cohort, median age was 63 years (IQR 53-70) with a male predominance (61%). Most had a DLBCL diagnosis (82%), ECOG 0-1 (89%), 2 prior lines of therapy (58%) and a high-risk CAR-HEMATOTOX (53%). Among patients with an available CNS-IPI score (442/875), 26% were low, 56% intermediate and 18% high-risk at time of CAR-T; 16% had received CNS prophylaxis with first-line treatment. Overall, the incidence of CNS relapse in CAR T-cell recipients was 1.6% (N=14). Focusing on these patients, 67% were germinal center B-cell (GCB) type and 29% were double/triple hit (DH/TH); all had received 2+ treatment lines and 21% prior CNS prophylaxis. Most (85%) had extranodal involvement and 21% had high-risk extranodal sites (breast, kidney, testes). The CNS-IPI score at CAR-T included 2 low, 7 intermediate and 4 high-risk (1 not available). Relapse was leptomeningeal (54%), parenchymal (23%) or both (23%), usually associated with concomitant systemic disease (71%). Median time from CAR T-cell infusion to PD was 1 month (IQR 0.9-3.5) and to CNS relapse was 4.7 months (IQR 1.4-48.1). At data cutoff, only 2 patients remained alive. Given the diverse settings, we explored a range of subgroup analyses. First, we focused on the 52% of patients who experienced PD after CAR-T, with a 3.1% incidence of CNS relapse. Second, we turned to CNS-IPI; in the 18% high-risk patients, 5.1% experienced CNS relapse. Third, we carried out an analysis restricted to patients with 1 prior treatment line (N=100), with no cases of CNS relapse. Regarding our comparative CIT cohort, baseline characteristics which were different from the CAR-T group included a younger age (57 vs 63 years, p<0.01), lower number of prior lines (1 in all patients), lower rate of transformed indolent lymphoma (6% vs 26%, p<0.01) and bulky disease (15% vs 23%, p=0.01). There were no differences in DH/TH status or CNS-IPI distribution; 26% had received prior CNS prophylaxis (p<0.01). The incidence of CNS relapse was comparable to the CAR-T cohort (2.2%, N=5 [p=0.60]). Of these, most (80%) had extranodal involvement, with high-risk sites (60%). The CNS-IPI score included 2 intermediate and 2 high-risk (1 not available). Relapse was mostly parenchymal (80%) and associated concomitant systemic disease (80%). Median time from CIT to PD was 1.7 months (IQR 1.2-4.1) and to CNS relapse was 3.5 months (IQR 2.1-8.7). None of these patients were alive at data cutoff. Finally, in the subgroup analyses, CNS relapse rate was 3.0% (p=1.0) among the 73% who experienced PD after CIT and 7.1% (p=0.7) in those with a high-risk CNS-IPI. Conclusions: The incidence of CNS relapse after CAR T-cell therapy was very low, especially in second-line. The CNS-IPI score did not seem to aid high-risk patient identification in this setting. Clinical trials incorporating CAR T-cell therapy in front-line will further clarify the role of CNS prophylaxis in this patient population.