IntroductionCD19-directed CAR T-cell therapy has transformed the therapeutic landscape for relapsed/refractory B-cell non-Hodgkin lymphoma (R/R B-NHL), offering durable responses in a subset of patients. However, durable remissions are seen in <40% of the patients, highlighting the need for predictive biomarkers to identify the likely responders and non-responders. Although post-infusion parameters such as CAR T-cell persistence, phenotype, and in vivo expansion, as well as tumor burden and immune microenvironment are known correlates of CAR T-cell efficacy, reliable baseline or pre-lymphodepletion biomarkers that could predict clinical response after CAR T-cell infusion remain largely undefined. Such biomarkers would be invaluable in guiding patient selection and treatment decisions. Decentralized CAR T-cell manufacturing offers a promising solution to enhance global accessibility and reduce costs, yet immune correlates of response and resistance from such real-world settings remain underexplored.

MethodsWe conducted a comprehensive immunophenotypic and proteomic analysis of peripheral blood mononuclear cells (PBMCs) and plasma samples from R/R B-NHL patients (n=27) enrolled in a phase I clinical trial using decentralized CD19 CAR T-cell manufacturing (Ghobadi et al., eClinicalMedicine 2025). Samples were collected at pre-lymphodepletion (day –6, referred henceforth as baseline), infusion (day 0), and serial post-infusion timepoints. Immune cells subset evaluation and T cell characterization such as differentiation, activation and exhaustion were done with 24-marker and 34-marker spectral flow cytometry panels respectively, while plasma protein levels were quantified using a 92-plex immuno-oncology panel (Olink®). Statistical comparisons were performed using Mann-Whitney and Kruskal-Wallis tests.

ResultsAmong the 27 patients, 26 were evaluated for response, and 1 patient died withing two weeks of infusion. Among these, 19 achieved complete remission and 3 patients achieved a partial remission as their best response in the first 6 months following infusion. Manual gating and unsupervised clustering of the T cells from the baseline samples showed a significantly higher frequency of early memory T cells (naive and central memory) in the responders, observed in both CD4+ and CD8+ populations. In contrast, non-responders had a significantly higher proportion of effector memory and terminal effector cells expressing KLRG1 and CD244. This difference was more pronounced in the CD4+ T cells compared to CD8+ T cells and was also identified at the early post-infusion (day 6) timepoints. Consistent with our earlier finding in B-cell acute lymphoblastic leukemia (Bai Z et al, Nature 2024), T cells (CD4+>CD8+) with type 2 function (CCR4+ non-regulatory T cells) were significantly enriched in the responders, a finding not described before in lymphoma. Regulatory T cells (Tregs) were notably reduced in non-responders, potentially reflecting their migration to lymphoid tissues, supported by elevated plasma CCL20 levels in non-responders.Proteomic analysis at baseline and early post-infusion timepoints revealed higher plasma concentrations of LAMP3 and ANGPT1 in responders and increased CCL20 in non-responders. Non-responders also exhibited elevated IL-6, GZMB, GZMH, CD40, and TNFRSF12A at baseline, indicating a pre-existing pro-inflammatory and immunosuppressive tumor microenvironment. Manual and unsupervised clustering of the other immune cells revealed a higher frequency of circulating monocytes (classical and non-classical) in the non-responders and a trend towards increased NK cells in the responders at baseline. Single-cell RNA sequencing from the baseline PBMCs is currently underway to further validate these findings.

ConclusionOur study demonstrates that distinct baseline immune and plasma protein profiles can predict response and resistance to CD19 CAR T-cell therapy in R/R B-NHL patients treated in a decentralized manufacturing setting. These findings highlight the importance of integrating baseline biomarker profiling into clinical workflows to improve patient selection, predict relapse, and optimize outcomes of CAR T-cell therapy in lymphoma.

This content is only available as a PDF.
2025
Sign in via your Institution