Secondary central nervous system (CNS) involvement with aggressive B-cell lymphoma (henceforth referred to as SCNSL), either at initial diagnosis, or at the time of disease relapse is associated with poor outcomes. Chimeric antigen receptor T-cell (CAR-T) therapy has revolutionized the treatment of non-Hodgkin lymphomas, however, most pivotal CAR-T trials excluded patients with SCNSL.¹ Little is known about the efficacy of CAR-T therapy in patients with history of versus active SCNSL. We report here the clinical outcomes of CAR-T in SCNSL patients, with focus on outcomes of patients with active versus prior secondary CNS involvement.

This is a retrospective study including SCNSL patients with a diagnosis of diffuse large-B-cell lymphoma (DLBCL), high-grade B-cell lymphoma, transformed follicular lymphoma (tFL), or Burkitt lymphoma who underwent commercial CD19-directed CAR-T therapies at 10 academic institutions in the United States. SCNSL involvement was defined as evidence of parenchymal or leptomeningeal involvement with lymphoma at any time prior to receiving CAR-T therapy. Active CNS disease was defined as the detection of lymphoma within the CNS at the last assessment performed prior to CAR-T infusion (magnetic resonance imaging and/or lumbar puncture).

Systemic and CNS responses were assessed using Lugano criteria² and the International Primary CNS Lymphoma Collaborative Group guidelines, respectively.³ Cytokine release syndrome (CRS) and immune cell-associated neurotoxicity syndrome (ICANS) were graded using the ASTCT consensus criteria.⁴ This study was approved by the institutional review boards at each participating institution. Survival analysis was done using Kaplan–Meier survival estimates and log-rank test.

A total of 113 CAR-T recipients with active (N = 86), or history of (N = 27) SCNSL were included in this analysis (Table 1). Median age at the time of CAR-T was 62 years. DLBCL was the most common subtype (N = 88; 78%) followed by tFL (N = 13; 12%). The median number of prior therapy lines was 3 (range: 1–8). Twenty-two patients (19%) underwent transplantation prior to CAR-T therapy. Nineteen patients (17%) received CNS-directed radiation therapy while 24 patients (21%) received a Bruton's tyrosine kinase inhibitor (BTKi) within a month prior to CAR-T infusion. Tisagenlecleucel (tisa-cel) was prescribed in 46 (41%) patients, axicabtagene ciloleucel (axi-cel) in 44 (39%), and liso-cel in 23 (20%) patients. Supplemental Table 1 shows baseline characteristics by the type of CAR-T product used. The median follow-up was 10.7 months (IQR 6.5, 30.69). Among the 86 (76%) patients with active CNS disease, the site(s) of CNS involvement included: parenchymal disease (N = 35; 41%), leptomeningeal involvement (N = 33; 38%), or both (N = 18; 21%).

CRS developed in 85 (75%) patients, and was mostly grade 1–2 (n = 81, 95%) (details in Supplemental Table 2). The median time to CRS onset was 3 days in patients with active CNS disease and 2 days in patients without active CNS disease. Tocilizumab was used in 55 patients. In patients with active CNS disease, CRS developed in 63 patients (73%), including grade 1–2 in 94%, grade 3 in 5%, and grade 5 in 1%. In patients without active CNS disease, CRS developed in 82% of patients, all of which were grade 1 or 2. Among all patients, ICANS developed in 63 patients (56%). The median time to onset of ICANS was 5 days in patients with active CNS disease and 6.5 days in patients without active CNS disease. Fifty-seven (50%) patients received steroids for management of ICANS. In patients with active CNS disease, any grade ICANS developed in 49 (57%) patients, with grade 1–2 developing in 24 (49%), grade 3 in 21 (43%), and grade 4 in 4 (8%) patients. In patients without active CNS disease, ICANs developed in 14 (52%) patients, including grade 1–2 in 57%, grade 3 in 36%, and grade 4 in 7%. No grade 5 ICANS was observed. In line with real-world experience, in the multivariable analysis (adjusting for CNS status), the use of liso-cel and tisa-cel were associated with decreased risk of ICANS [Odds Ratios (OR), 0.21(95% CI = 0.067–0.62, p < .01) and 0.17 (95% CI = 0.06–0.44, p < .001), respectively]. The presence of active CNS disease and CNS radiation within 1 month of CAR-T infusion was not associated with a risk of developing ICANS (Supplemental Table 3).

Among patients evaluable for CNS response (N = 80), the overall response rate (ORR) was 68% at 1 month with a CNS complete response (CR) rate of 34%. The 1-month ORR for patients with leptomeningeal disease was 67% compared to 72% in patients with parenchymal disease and 59% in patients with both parenchymal and leptomeningeal disease.

Among patients evaluable for systemic response at 1 month (N = 93), the ORR was 75%, with 58% of patients achieving a CR. In patients with active CNS disease, the systemic ORR at 1 month was evaluable in 68 patients and was 76%. Supplemental Tables 4 and 5 show responses at 1 and 3 months by active CNS disease and by sites of active CNS disease. Cumulative incidence of CNS or systemic relapse for the total patient population at 12 months was 69% (95% CI = 61%–79%). In patients with active CNS disease, cumulative incidence of relapse at 12 months was 78% (95% CI = 69%–89%).

Among patients with leptomeningeal disease at the time of CAR-T (n = 25), relapse manifested as leptomeningeal disease in 29%, followed by both leptomeningeal and systemic in 25%, then systemic disease only in 21%. Among patients with parenchymal disease at the time of CAR-T (n = 23), relapse was parenchymal in 36%, followed by both parenchymal and systemic in 23%, and systemic only in 23% of patients. Among patients who had both parenchymal and leptomeningeal disease at the time of CAR-T, relapse was both leptomeningeal and parenchymal in 27% of cases, followed by parenchymal only in 20%, and ocular in 20%.

In patients with active CNS disease, the 12-month progression-free survival (PFS) was 17% (95% CI = 9.7%–29%) and the median PFS was 2.9 months (95% CI = 2–3.8). The 12-month overall survival (OS) in patients with active CNS disease was 39% (95% CI = 29%–52%), and the median OS was 8.6 months (95% CI = 5.3–13). In patients without active CNS disease, the 12-month PFS was 53% (95% CI = 37%–77%). The median PFS was 14 months (95% CI = 6-not reached). The 12-month OS for patients without active CNS disease was 77% (95% CI = 63%–95%) and the median OS was not reached (Supplemental Figures 1–3).

In patients with active versus history of CNS disease, non-relapse mortality at 12 months was 5.1% (95% CI = 2%–13%) versus 3.7%(95% CI = 0.5%–25%), respectively (Supplemental Figure 4).

In the multivariable model for PFS, the presence of active CNS disease was associated with inferior PFS with (Hazard Ratio [HR] = 1.9 [95% CI = 1.17–3.08, p = .009]). Liso-cel was associated with significantly improved PFS (HR = 0.48 [95% CI = 0.27–0.84, p = .01]) (Supplemental Table 6). In the multivariable model for OS, the presence of active CNS disease was associated with significantly inferior OS with (HR = 2.34 [95% CI = 1.26–4.34, p = .007]). Use of BTKi or CNS-directed radiation within 1 month of CAR-T infusion was not associated with OS (HR = 1.24 [95% CI = 0.85–1.79] and 1 [95% CI = 0.79–1.27] respectively). The use of different types of CAR-T products was not associated with a significant difference in OS (Supplemental Table 7). Supplemental Figure 5 shows PFS and OS by type of CAR-T product used.

This large retrospective analysis demonstrates the feasibility and safety of commercial CAR-T in patients with SCNSL. Despite the initial responses to CAR-T, patients with active CNS disease had significantly inferior PFS and OS compared to patients with history of, but no active SCNSL at the time of CAR-T infusion. This demonstrates the limited durability of CAR-T responses in SCNSL patients with active CNS disease consistent with recently reported data.⁵ Notably, our work emphasizes that strategies to improve outcomes with CAR-T therapy, such as the use of radiation or BTKi prior to CAR-T, warrant further investigation. In our analysis, neither the use of BTKi nor radiation had a significant impact on OS in the multivariable models, however, our analysis was limited by the relatively short timeframe of exposure to BTKi, and only a limited number of patients who received radiation within a month of CAR-T (n = 19). However, we show that radiation was not associated with increased ICANS, demonstrating the feasibility of this as a bridging strategy in patients with SCNSL.

The incidence of grade 3 or higher ICANS in our study is comparable to the rate of ICANS reported in the literature.¹ The type of CAR-T product used was more likely to impact the incidence of ICANS, with axi-cel being associated with higher rates when compared to tisa-cel and liso-cel. Notably, we did not observe a difference in rates of ICANS in patients with versus without active CNS disease. We were unable to attribute neurological toxicities to tumor inflammation-associated neurotoxicity (TIAN) versus ICANS as TIAN was not defined at the time of patient treatment. TIAN is caused by neuronal dysfunction due to the inflammation-induced tumoral edema or neuro-immune interactions as an on target-treatment effect.⁶ This is in contrast to cerebral edema observed in ICANS. Future studies should attempt to characterize the subtle distinction between, and potential overlap of TIAN and ICANS in SCNSL. Our study was limited by its retrospective nature and the heterogeneity in provider practice, patient population, and CAR-T products used across US centers. Despite these limitations, our findings promote the inclusion of patients with SCNSL in clinical trials evaluating the efficacy of therapeutic strategies that include CAR-T therapy.

AUTHOR CONTRIBUTIONS

AER, GC, PAR, FA, LS, MS, MH, VB, WW, NA, MI, MKD, MS, PCJ, YC, and SI performed research, collected data, interpreted data, and reviewed manuscript. AS and JW performed the statistical analysis. MH and MJF designed research, performed the research, interpreted the data, drafted, and reviewed the manuscript. GA designed research, performed the research, collected data, interpreted data, and reviewed the manuscript. AA designed research, performed the research, collected data, interpreted data, drafted and reviewed the manuscript. All the authors reviewed and approved the submitted manuscript.

CONFLICT OF INTEREST STATEMENT

MJF: Consulting/research support from BMS, Arcellx, Kite/Gilead, JnJ/Legend, Novartis, and Cytoagents. MH: research support/Funding: Takeda Pharmaceutical Company; ADC Therapeutics; Spectrum Pharmaceuticals; Astellas Pharma. Consultancy: ADC Therapeutics, Omeros, CRISPR, BMS, Kite, Abbvie, Caribou, Genmab. Speaker's Bureau: ADC Therapeutics, AstraZeneca, BeiGene, Kite. DMC: Inc, Genentech, Myeloid Therapeutics, CRISPR. NA: Ad Board: Bristol Myers Squib. Institutional Research Funding and consultancy: Kite/Gilead. MKD: Research/grant: from Novartis, Bristol Myers Squibb and Pharmacyclics. Consultancy: Kite Pharma. MS: served as a paid consultant for McKinsey & Company, Angiocrine Bioscience, Inc., and Omeros Corporation; received research funding from Angiocrine Bioscience, Inc., Omeros Corporation, and Amgen, Inc.; served on ad hoc advisory boards for Kite-A Gilead Company; and received honoraria from i3Health, Medscape, and CancerNetwork for CME-related activity. VB: serves on DSMB for Miltenyi Biotech, is member of ad hoc advisory board for Astra Zeneca, ADC, Allogene and BMS, received research funding from Incyte, Gamida Cell, and Citius. PR: has served as a consultant and/or advisory board member for AbbVie, Novartis, BMS, ADC Therapeutics, Kite/Gilead, Sana Biotechnology, Nektar Therapeutics, NurixTherapeutics, Intellia Therapeutics, CVS Caremark, Genmab, BeiGene, Janssen, and Pharmacyclics. He has received honoraria from Novartis. Research support from BMS, Kite Pharma, Novartis, MorphoSys, CRISPR Therapeutics, Calibr, Xencor, Fate Therapeutics, AstraZeneca, Genentech, and Tessa Therapeutics. PCJ: reports Consulting for AstraZeneca, Abbvie, ADC Therapeutics, Bristol Myers Squibb, Incyte, and Seagen and Research Funding from AstraZeneca, Incyte, and Medically Home.

INFORMED CONSENT

After review by the institutional review boards at each of the participating institutions, informed consent was waived given the retrospective nature of this study.