henhaobio
Sign in

Cell ReportsImmunologyPeer-reviewed journal article

A balanced CAR-T cell metabolism driven by CD28 or 4-1BB co-stimulation correlates with clinical success in lymphoma patients

Molly S. Cook, Ellen King, Katie R FlahertyiD, Khadiza Siddika, Sophie PapaiD, Reuben BenjaminiD, Anna SchurichiD

PDFDOIPreprint
In-depth full-text digest available · Jump to digest ↓

Abstract

Chimeric antigen receptor (CAR) T cell therapy has led to unprecedented success in treating relapsed/refractory diffuse large B cell lymphoma (DLBCL). The most common CAR-T cell products currently in the clinic for DLBCL differ in their co-stimulation moiety, containing either CD28 or 4-1BB, which initiate distinct signalling pathways. Previous work has highlighted the importance of T cell metabolism in fuelling anti-cancer function. We have studied the metabolic characteristics induced by CD28 versus 4-1BB co-stimulation in patient CAR-T cells ex vivo. Our data show that in patients, CD28 and 4-1BB drive significantly divergent metabolic profiles. CD28 signalling endows T cells with a preferentially glycolytic metabolism supporting an effector phenotype and increased expansion capacity, while 4-1BB co-stimulation preserves mitochondrial fitness and results in memory-like differentiation. Despite this divergent programming, T cells in patients responding successfully to therapy were metabolically similar, irrespective of co-stimulator, suggesting that efficient fuelling of CAR-T cells in lymphoma requires a balanced metabolism. In contrast, CAR-T cells in non-responders were pushed to their metabolic extremes. One sentence summaryCD28 and 4-1BB signalling drive divergent metabolic profiles in patient CAR-T cells, however patients that respond to therapy have CAR-T cells that maintain metabolic balance.

In-depth summary

Claude Sonnet 4.6 · xhigh effort · 8 Nov 2026 · from the full text of version 1 (29 Sept 2024) · CC BY 4.0

Ex vivo profiling of patient CD19 CAR-T cells shows CD28 products skew glycolytic and effector-like while 4-1BB products favor OXPHOS; complete responders show balanced metabolism.

Around 50% of DLBCL patients fail CAR-T therapy, and CAR-T metabolism in patients is hard to study because circulating cell numbers are low. This work compares CD28 and 4-1BB products at single-cell level ex vivo in patients and relates metabolic balance to 6-month response.

Key results

  1. ShownCD28 CAR-T cells had higher glycolysis-linked protein expression (GLUT-1, phosphorylated mTOR at Ser2448, CD98) than 4-1BB CAR-T cells. PGC1a, CD71 and CPT1A did not differ.

    Flow cytometry of ex vivo CAR-T at day 7 post infusion in 14 CD28 (axi-cel) vs 9 4-1BB (tisa-cel) patients; Mann Whitney U test, P<0.05 threshold; GLUT-1, p-mTOR and CD98 significantly higher with CD28. · CD28 and 4-1BB signalling differentially…; STATISTICS; Figure legends

  2. ShownSCENITH on CD69+ CAR-T cells showed 4-1BB products depend more on OXPHOS and CD28 products more on glycolysis, despite similar total translation. CD28 cells had more CD38+ cells; polarised mitochondria trended higher in 4-1BB cells.

    Puromycin incorporation (SCENITH) with oligomycin-A and 2-deoxy-D-glucose; 14 CD28 vs 9 4-1BB patients; Mann Whitney U test; total translation similar between products. · Glycolysis fuels protein translation in…; Discussion; Metabolic flux analysis (SCENITH); Figure legends

  3. ShownCD28 CAR-T cells had a more effector-like phenotype (more CD69+, higher PD1, more CD57+ cells), while 4-1BB cells had a memory-like phenotype with more CD62L+ cells. CD103 and CD39 did not differ significantly.

    Flow cytometry at day 7 in 14 CD28 vs 9 4-1BB patients; Mann Whitney U test; CD57 increase described as striking with CD28, CD62L increased with 4-1BB. · The ex vivo CAR-T metabolic profile is…; Figure legends

  4. SuggestedMetabolic and phenotypic differences between CD28 and 4-1BB products were clearer mainly in non-responders. PCA showed greater overlap of the two products in complete responders, which supports balanced metabolism as a correlate of response.

    Responders (complete response) vs non-responders (progressive disease) at 6 months, Lugano 2014; GLUT-1 and polarised mitochondria differed between products only in non-responders; subgroup sizes not stated. · A balanced CAR-T cell metabolism correlates with…; Figure legends

  5. SuggestedCD28 recipients had a higher frequency of circulating CAR+ T cells at day 7 and a non-significant trend toward better overall and progression-free survival.

    CAR+ frequency at day 7 higher with CD28 (Mann Whitney U test, P<0.05); Kaplan-Meier trend only in this 14 vs 9 patient cohort, not significant. · Increased frequency of circulating CD28…; Discussion; Figure legends

Methods, data and reproducibility

Cryopreserved PBMCs from DLBCL patients 7 days after CAR-T infusion were profiled by multiparameter flow cytometry, mitochondrial dyes and SCENITH, with CytoNorm batch normalisation. CD28 and 4-1BB products were compared by Mann Whitney U test and PCA, stratified by 6-month response.

  • Ex vivo PBMCs from DLBCL patients given CD28 (axicabtagene ciloleucel) or 4-1BB (tisagenlecleucel) CD19 CAR-T, sampled at day 7 post infusion · human · n = 23 patients (14 CD28, 9 4-1BB)
  • Single healthy-donor CD19 CAR-T control, cryopreserved in 40 vials and run with each phenotyping batch for CytoNorm normalisation · human · n = 1 donor, 40 vials
Methods & tools
Flow cytometry (BD Fortessa), SCENITH, MitoTracker Deep Red / Green polarisation assay, CytoNorm (R), FlowJo v10, GraphPad Prism 9, PCA, Mann Whitney U test, Lugano criteria 2014
Reagents & environment
FlowJo v10; GraphPad Prism 9; CytoNorm package in R (version not stated).
Reproducibility gaps
No original code reported; data available only on request from the lead contact; patient samples come from a biobank under ethics approval.
Limitations (4)
  • The cohort is small (14 CD28, 9 4-1BB) and prior treatment, ethnicity, sex, age and disease burden could not be controlled; 4-1BB patients were significantly older. · stated by the authors
  • Survival differences between CD28 and 4-1BB recipients were non-significant trends, so clinical conclusions need cautious interpretation. · stated by the authors
  • Responder versus non-responder comparisons rest on subgroups of an already small cohort; subgroup sizes and effect sizes are not reported, and the 'balance' conclusion rests on loss of significance within subgroups, which is not a formal interaction test. · noted in this summary
  • Metabolism was measured at a single time point (day 7) on circulating cells only; no functional or causal test links balanced metabolism to efficacy, and tumour-resident CAR-T were not examined. · noted in this summary

Read this if you work on CAR-T metabolism or biomarkers and want ex vivo patient data comparing CD28 and 4-1BB products in DLBCL.

Citations
55
FWCI
23.37
Field percentile
top 1%

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISGoldberg '26Peeters '26Lakhani '23Cao '25Tang '26Gross '23
71% semantic matchImmunology
Lior Goldberg, Eric R. Haas, Jiaqi Wu+25 more·Nature Communications·Feb 20, 2026· 3 citations