Introduction

Chimeric antigen receptor T cell (CAR T) therapy for relapsed/refractory large B-cell lymphoma (LBCL) achieves long-term remission in only about 30-40% of patients. Prior studies suggest that T cell subset composition and functionality are associated with CAR T outcomes. While T cells are known to rely on distinct metabolic pathways for differentiation and function, limited data exists on how metabolic activation potential influences early CAR T failure.

Methods Peripheral blood mononuclear cells (PBMCs) were collected via leukapheresis from pts with LBCL prior to CAR-T and healthy donors (HD). PBMCs were analyzed with and without 72-hour CD3/CD28 stimulation using mass cytometry by time of flight (CyTOF) to assess T cell activation and subset composition. A CyTOF panel of 26 key proteins spanning 7 metabolic pathways was used to evaluate the metabolic response. The selected metabolic markers covered 7 pathways, including transcription factors (HIF1α, KEAP1, p-PCG1a ), fatty acid metabolism (p-ACC, CD36, CPT1A, ACADM), tricarboxylic acid cycle (TCA) (CS, IDH1), amino acid metabolism (GLS, CD98, GLUD12, p-S6), glycolytic pathway enzymes (PFKFB4, GLUT1, LDHA, HK2, GAPDH, PDK1), mitochondrial metabolism (VDAC1, CytC, ATP5A), and pentose phosphate pathway (G6PD).

High-dimensional analysis was performed using standard CyTOF workflows. The cyCombine algorithm was applied to integrate single-cell datasets from resting and stimulated samples, using a healthy donor reference sample across batches for normalization. Metabolic scores were calculated as the difference in summed marker expression between unstimulated and stimulated conditions for each pathway, and compared using Wilcoxon rank sum test.

Results Single-cell analysis included 22 patients (21 axi-cel, 1 tisa-cel; median age 62 years [range 39-76]). Diagnoses included DLBCL (n=14), HGBL (n=5), transformed lymphoma (n=2), and PMBCL (n=1). Most patients were refractory to their last treatment (64%) and had received a median of 2 prior lines of therapy. High-risk features included IPI scores of 3–5 (41%), ECOG ≥2 (23%), stage III-IV disease (64%), elevated LDH (median 281 U/L), and ≥2 extranodal sites (18%). At 3-month post CAR T, 15/22 patients achieved complete response (CR), 4/22 had partial response (PR), and 3/22 had progressive disease (PD).

Metabolic activation scores were consistently higher in patients with CR and PR compared to PD across most pathways in both CD8⁺ and CD4⁺ T cells.

Amino acid metabolism scores were significantly higher in patients with early CR (CD8⁺: 0.54 ± 0.36, CD4⁺: 0.73 ± 0.59) compared to early PD (CD8⁺: 0.04 ± 0.09, CD4⁺: 0.09 ± 0.16, p<0.001), while PR was similar to CR (CD8⁺: 0.52 ± 0.22, CD4⁺: 0.66 ± 0.41).

Similar patterns were observed in the following pathway scores:

Glycolysis (CD8⁺ CR 0.38 ± 0.32, PR 0.37 ± 0.21, PD -0.06 ± 0.13, p<0.05; CD4⁺ CR 0.54 ± 0.41, PR 0.63 ± 0.36, PD ≈0.00 ± 0.07, p<0.001);

Mitochondrial dynamics (CD8⁺ CR 0.41 ± 0.38, PR 0.45 ± 0.30, PD -0.20 ± 0.26, p<0.05; CD4⁺ CR 0.70 ± 0.72, PR 0.67 ± 0.87, PD -0.30 ± 0.26, p<0.001);

Pentose phosphate pathway (CD8⁺ CR 0.33 ± 0.49, PR 0.52 ± 0.26, PD -0.02 ± 0.11; CD4⁺ CR 0.51 ± 0.34, PR 0.73 ± 0.14, PD 0.07 ± 0.18, p<0.05);

TCA (CD8⁺ CR 0.45 ± 0.31, PR 0.52 ± 0.19, PD 0.06 ± 0.15; CD4⁺ CR 0.58 ± 0.33, PR 0.74 ± 0.21, PD 0.12 ± 0.11);

Fatty acid metabolism (CD8⁺ CR 0.41 ± 0.37, PR 0.55 ± 0.34, PD -0.23 ± 0.56; CD4⁺ CR 0.52 ± 0.42, PR 0.61 ± 0.39, PD -0.32 ± 0.36, p<0.001);

Transcriptional pathway (CD8⁺ CR 0.46 ± 0.37, PR 0.40 ± 0.17, PD -0.05 ± 0.15, p<0.05; CD4⁺ CR 0.72 ± 0.63, PR 0.61 ± 0.50, PD -0.09 ± 0.12, p<0.001).

Conclusion Patients with early PD following CAR-T therapy demonstrated consistently reduced metabolic activation across multiple metabolic pathways in CD8⁺ and CD4⁺ T cell subsets in pre-infusion PBMCs, suggesting impaired metabolic fitness may underlie treatment resistance. These findings support development of an ImmunoFit score to quantify T cell metabolic readiness and inform strategies to enhance CAR T efficacy, such as metabolic modulation prior to infusion. Further validation is ongoing, but this framework offers a biological approach to improving clinical outcomes.

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