In-depth, quantitative characterization of important regulators and predictive biomarkers for CAR T-cell therapy induced toxicities in a clinical cohort

Abstract

Abstract Background: CAR T cells targeting B cell maturation antigen (BCMA) and CD19 achieve high response rates in B cell malignancies.However, toxicity remains a significant clinical challenge, with adverse events such as immune effector cell-associated neurotoxicity syndrome (ICANS) frequently observed following CAR T-cell infusion. To better address this toxicity and improve patient management, we aimed to investigate the mechanisms of ICANS by studying the dynamic changes of proteins in plasma and immune cells. In addition, we sought to identify candidate predictive biomarkers for CAR T-induced toxicities by comparing the pre-treatment protein profiles of patients who later developed ICANS with those who did not. Methods: We analyzed a longitudinal cohort of 38 patients with relapsed/refractory multiple myeloma (n=21), non-Hodgkin lymphoma (n=16), and plasma cell leukemia (n=1) who underwent standard of care CAR-T therapy. We collected over 100 blood samples across 16 time points from Day 0 to Day 30 post-treatment and plasma and immune cells, including neutrophils, monocytes, and T cells, were isolated for proteomic profiling. The samples were prepared using a reproducible protocol and analyzed with a highly sensitive and robust LC-MS pipeline, featuring a trapping nano-LC system and Orbitrap Astral mass spectrometer. We quantified 4,000–6,000 proteins using nanogram-level inputs with high precision (e.g., median protein CV = 12.7%). Time-course bioinformatics analyses were generated for over 2,500 plasma proteins and more than 6,000 immune cell proteins per patient, with protein expression normalized by subtracting log2 intensity at Day 0. Patients with no toxicity were compared to those who developed ICANS to identify proteins associated with onset and progression. Additionally, baseline (day 0; pre-treatment) profiles were examined to identify potential predictive biomarkers for ICANS. Proteins that are significantly increased at baseline (fold change >1.4-fold, p < 0.05) in patients who later developed ICANS are candidate predictive biomarkers for ICANS. Results: Baseline comparisons of plasma from control patients (no ICANS) and those who went on to develop ICANS (grades 1-5) revealed candidate biomarkers associated with ICANS development. These proteins are enriched in pathways related to glutathione metabolism and hydrogen peroxide catabolism (e.g., GLRX, GSR, GSHR), and MHC class I antigen presentation (e.g., HLA-A), suggesting a role for oxidative stress, redox regulation, and adaptive immunity in ICANS susceptibility and identifying potential avenues for early intervention. Baseline comparisons of neutrophils showed that proteins involved in complement activation (e.g., C4B, CFB, CLU, C9, C3, C7), acute-phase response (e.g., SERPINA1, SERPINF2), and platelet activation (e.g., VCL, FGA) were increased at baseline in patients who later developed ICANS. Longitudinal analysis of plasma proteins revealed distinct expression clusters, with upregulation of proteins involved in leukocyte and platelet aggregation as well as complement proteins over time in patients with ICANS. In these patients, neutrophils exhibited dynamic regulation of protein complex assembly and wound healing, highlighting their role in inflammation. T cells showed temporal changes in RNA splicing-related proteins consistent with T cell activation. Monocytes showed time-dependent regulation of vesicle organization and transport, suggesting involvement in antigen presentation and cytokine secretion. Collectively, these data suggest that complex immune crosstalk underlies the pathophysiology of ICANS. Conclusions: This is the first in-depth, time-resolved proteomic profiling of plasma and immune cells during CAR-T therapy, offering new insights into ICANS mechanisms that can guide risk stratification and management strategies for patients with multiple myeloma and lymphoma.