Naive CD4+ at apheresis and disease control at infusion are associated with improved efficacy in second-line CAR T-cells

Abstract

Abstract Introduction: Second-line (2L) anti-CD19 CAR T-cell therapy (CART) with axicabtagene ciloleucel (axi) or lisocabtagene maraleucel (liso) is standard-of-care for patients (pts) with large B-cell lymphomas (LBCLs) who relapse within 12 months after frontline treatment. However, determinants of response and toxicity in 2L real-world settings remain poorly defined. Methods: We retrospectively analyzed 64 consecutive LBCLs pts treated at our institution with 2L axi (n=35) or liso (n=29) between 05/2022 and 12/2024. Multiparameter spectral immunophenotyping (Citek) was performed on peripheral blood collected at apheresis and day 7 post-CART. The primary objectives were to compare the two products in a real-world setting and to identify predictors of efficacy and toxicity. Results: Median age at time of CART was 62 years (21-80yrs); 59% had primary refractory LBCL, 17% had an HGBL, 66% had III/IV-stage and 38% had elevated LDH at infusion. Median vein-to-vein time was 42 days (median axi=36 vs liso=43; p=0.03). Bendamustine was used for lymphodepletion in 89% (axi=83% vs liso=96%, p=0.1). Overall and complete response rates (ORR and CR) at day 90 were 66% and 55%.At a median follow-up of 17 months (4-34mo), 12-month PFS and OS rates were 47% and 78%. Efficacy was comparable between axi and liso (12-month PFS: 51% vs 48%, p=0.8; best CR: 51% vs 58%, p=0.6), despite older age (mean 56 vs 66yrs, p<0.01) and higher IPI (≥3 in 28% vs 72%, p<0.01) in the liso cohort.Cytokine release syndrome (CRS) of any grade occurred in 52% of pts (G≥3 = 3%) and neurotoxicity (ICANS) in 12% (G≥3 = 3%). No non-relapse mortality or secondary malignancies occurred. Axi was associated with higher toxicity, with increased rates of CRS (any grade: 80% vs 17%, p<0.01) and a trend towards higher ICANS (any grade: 20% vs 3%, p=0.06). Due to few severe CRS/ICANS events, comparisons between axi and liso were not performed.Disease reassessment after bridging therapy was available for 92% of pts and 34% were infused in progression. Pts infused with progressive disease had worse outcomes as compared to those with disease control: CR as best response in 15% vs 73% (OR=0.06 95%CI=0.01-0.3, p<0.01), 12-month OS 48% vs 91% (p<0.01), and PFS 21% vs 59% (p<0.01). Bulky disease, extranodal involvement, and elevated LDH at infusion were negative predictors of durable response (defined as PFS≥12 months); interestingly these features were not prognostic when analyzed at the relapse after the frontline therapy.Given the prognostic value of clinical variables at infusion but not at relapse, we explored flow cytometry to identify early predictors of treatment response. Higher CD3⁺ and CD4⁺ T-cell counts at apheresis predicted PFS and CR to CART. Optimal cutoffs (401 CD3⁺/µL, 201 CD4⁺/µL) stratified pts with significantly prolonged 12-month PFS rates (59% vs 29% for CD4⁺; p<0.01). High Naive CD4⁺ T-cell at apheresis strongly correlated with durable responses and prolonged PFS (12-month PFS rates: 12% vs 87%, p<0.01).At day 7, responders showed a trend toward greater CART expansion (CAR⁺/CD3⁺: 17% vs 6%, p=0.1) and higher frequencies of naive CD8⁺CART. In contrast, pts infused with progressive disease had numerically lower CART expansion and a more differentiated effector memory phenotype (CAR+/CD3+: 5% vs 14%, p=0.1 and CD45RA-/CCR7-: 54% vs 38%, p=0.3). CAR⁻ bystander T-cells showed a more naive CD4⁺ and less effector CD8⁺ profile indicating functional divergence from CART.In multivariate analysis including CD4⁺ count at apheresis, disease progression at infusion, and primary refractory status, higher CD4⁺ count remained independently associated with higher odds of CR (OR=4.31, 95%CI=1.15-18.8, p=0.03). Progressive disease at infusion (OR=0.06, 95%CI=0.01-0.3, p<0.01) and primary refractory disease(OR=0.2, 95%CI=0.05-0.8, p=0.03) were associated with lower odds of CR. Conclusions: This is the first real-world study to define clinical and immune correlates of efficacy in 2L CART therapy for LBCL. Both axi and liso are safe and effective. Naïve CD4⁺ T-cell abundance at apheresis predicts long-term outcomes and may serve as a biomarker of T-cell fitness and CART potency. Progressive disease at the time of infusion is associated with inferior response rates and survival outcomes, which may be due to less favorable expansion kinetics and T-cell phenotypes. These findings support immune profiling at apheresis and rational bridging to improve outcomes in early-line CART therapy.