P1458: PATTERNS OF LONG-TERM HEMATOPOIETIC RECOVERY IMPACT SURVIVAL OUTCOMES AFTER CD19-DIRECTED CAR T-CELL THERAPY FOR R/R LBCL

Abstract

Background: Real-world evidence has underlined the role of hematological toxicity and infections in driving the toxicity burden of CD19 CAR-T. We have recently developed a classification system for post-CAR-T hematotoxicity, identifying three unique phenotypes of neutrophil recovery (Rejeski et al Blood 2021). However, their prognostic impact remains poorly understood. Aims: We sought to understand potential pathomechanisms underlying hematopoietic recovery patterns, and characterize their influence on toxicity and clinical outcomes. Methods: In this multicenter retrospective observational study, we analyzed 316 pts. receiving Axicabtagene ciloleucel (n=194) or Tisagenlecleucel (n=122) for R/R LBCL in a real-world setting. The phenotypes of neutrophil recovery were defined as follows: 1) Quick Recovery: sustained neutrophil recovery without a second dip below an ANC<1000/µl. 2) Intermittent Recovery: neutrophil recovery (ANC>1500/µl) followed by a second dip with an ANC<1000/µl after day 21. 3) Aplastic: continuous severe neutropenia (ANC<500/µl) ≥14 days. Multivariate analysis was performed as stepwise binary logistic regression for the aplastic vs. non-aplastic phenotypes. Variables with a p<0.2 on univariate analysis were included. Clinical outcomes were studied via Kaplan-Meier estimates. Serum cytokines were analyzed using the Ella automated immunoassay system (ProteinSimple) in 70 pt. from Moffitt Cancer Center. The plasma proteome was characterized in a further 38 pts. from the LMU Munich across four time points (day 0, 4, 14, 28) using a 92-protein multiplex proximity extension assay (Olink Bioscience). Results: The overall distribution of the ‘quick’ (Q), ‘intermittent’ (I) and ‘aplastic’ (A) phenotypes was 40%, 41%, and 19%, respectively. Prior to lymphodepletion, ‘aplastic’ patients displayed increased systemic inflammation (median ferritin 1168 ng/ml), impaired hematopoietic function (median platelet count 101 G/l) and high tumor burden (median LDH 344 U/l). ‘Aplastic’ patients displayed a longer median duration of severe neutropenia, (A vs. I vs. Q: 28 vs. 10 vs. 5 days, p<0.001), higher rate of severe infections (A vs. I vs. Q: 40% vs. 22% vs. 18%), and a trend towards higher 1-year non-relapse mortality (A vs. I: 10% vs. 4%, p=0.09). Progression-free (PFS) and overall survival (OS) was poor in the ‘aplastic’ patients (Fig. 1, 1-yr PFS 24%, 1-yr OS 48%). On the other hand, ‘intermittent’ patients exhibited the best survival outcomes (HR for PFS, I vs. A: 0.46, 95% CI 0.32-0.69). Median PFS was 3 vs. 4 vs. 20 months (A vs. Q v. I, p<0.001); median OS was 8 vs. 23 months vs. not reached (A vs. Q v. I, p=0.0025). On multivariate regression (n=316), the aplastic phenotype was primarily driven by baseline ECOG (OR=1.6, p=0.05), ferritin (OR=2.5, p=0.01), ANC (OR=0.3, p=0.01), and hemoglobin (OR=0.8, p=0.04). Of note, high-grade CRS/ICANS and CAR product were not associated with the aplastic phenotype. Initial analyses of inflammatory patterns revealed higher peak IL-15 levels and an increased ratio of ANG2 to ANG1 in ‘aplastic’ patients in both cohorts, with ANG-1 levels decreasing over time. Image:Summary/Conclusion: Hematopoietic recovery patterns impact disease outcomes after CD19 CAR-T. Surprisingly, a phenotype associated with recurrent neutrophil dips (i.e. prolonged cytopenia) was associated with improved survival outcomes. The poor-risk ‘aplastic’ phenotype was driven by baseline inflammation and hematopoietic function, and characterized by progressive endothelial dysfunction. Further mechanistic exploration is warranted and ongoing.