AFFILIATIONS
1 University of Pennsylvania, Perelman School of Medicine, Philadelphia, Pennsylvania, United States
2 University of Pennsylvania, Philadelphia, Pennsylvania, United States, University of Pennsylvania, Philadelphia, Pennsylvania, United States
3 University of Pennsylvania, Philadelphia, Pennsylvania, United States
4 UNSW, Sydney, Australia, Westmead Institute for Medical Research, Westmead, Australia
5 Childrens Hospital of Philadelphia, Philadelphia, Pennsylvania, United States
6 University of Pennsylvania, Philadelphia, Pennsylvania, United States, University of Pennsylvania, Philadelphia, Pennsylvania, United States
7 Childrens Hospital of Philadelphia, Philadelphia, Pennsylvania, United States
8 Childrens Hospital of Philadelphia, Philadelphia, Pennsylvania, United States
9 University of Pennsylvania, Philadelphia, Pennsylvania, United States, University of Pennsylvania, Philadelphia, Pennsylvania, United States
10 University of Pennsylvania Perelman School of Medicine, Philadelphia, Pennsylvania, United States
11 University of Pennsylvania Perelman School of Medicine, Philadelphia, Pennsylvania, United States
12 University of Pennsylvania Perelman School of Medicine, Philadelphia, Pennsylvania, United States
13 Brigham and Womens Hospital, Boston, Massachusetts, United States, University of Pennsylvania Perelman School of Medicine, Philadelphia, Pennsylvania, United States, University of Pennsylvania Perelman School of Medicine, Philadelphia, Pennsylvania, United States
14 University of Pennsylvania, Philadelphia, Pennsylvania, United States
15 University of Pennsylvania, Philadelphia, Pennsylvania, United States
16 University of Pennsylvania, Philadelphia, Pennsylvania, United States
17 University of Pennsylvania, Philadelphia, Pennsylvania, United States
18 University of Pennsylvania, Philadelphia, Pennsylvania, United States
19 University of Pennsylvania, Philadelphia, Pennsylvania, United States
20 University of Pennsylvania Perelman School of Medicine, Philadelphia, Pennsylvania, United States
21 University of Pennsylvania, Philadelphia, Pennsylvania, United States, University of Pennsylvania, Philadelphia, Pennsylvania, United States
22 University of Pennsylvania, Philadelphia, Pennsylvania, United States, University of Pennsylvania, Philadelphia, Pennsylvania, United States
23 University of Pennsylvania Perelman School of Medicine, Philadelphia, Pennsylvania, United States, University of Pennsylvania, Philadelphia, Pennsylvania, United States
24 University of Pennsylvania, Philadelphia, Pennsylvania, United States, University of Pennsylvania, Philadelphia, Pennsylvania, United States
25 UNSW, Sydney, Australia, Westmead Institute for Medical Research, Westmead, Australia
26 University of Pennsylvania, Philadelphia, Pennsylvania, United States, University of Pennsylvania, Philadelphia, Pennsylvania, United States, University of Pennsylvania, Philadelphia, Pennsylvania, United States
27 Childrens Hospital of Philadelphia, Philadelphia, Pennsylvania, United States, Childrens Hospital of Philadelphia, Philadelphia, Pennsylvania, United States, University of Pennsylvania Perelman School of Medicine, Philadelphia, Pennsylvania, United States
ABSTRACT
Background: CD19-directed chimeric antigen receptor T-cell therapy (CART19) is a standard treatment for relapsed/refractory large B-cell lymphoma (r/r LBCL), but many patients do not achieve durable remission. Emerging evidence implicates myeloid cells in regulating CAR T-cell function; however, longitudinal myeloid–CAR T-cell interactions remain poorly defined.
Methods: We conducted a longitudinal study of 15 selected patients with r/r LBCL treated with CART19 (tisa-cel/CTL019) with extreme divergent phenotypes: 8 with complete response (CR) lasting >1 year and 7 with progressive disease (PD) at first follow up. We profiled 67 samples comprising peripheral blood mononuclear cells (PBMCs) collected at apheresis (n=15), day 0 before infusion (n=15), peak CAR T-cell expansion (days 7-14) (n=13), and day 28 (n=9), plus matched infusion products (IPs; n=15), by CITE-seq (49 antibodies). We also performed co-detection by indexing (CODEX; 54 markers) on 10 matched pretreatment biopsies (5 CR and 5 PD). We compared gene expression between response groups, performed gene set enrichment analysis (GSEA), inferred cell–cell communication (CCC) with LIANA, modeled longitudinal cell-state transitions with MOSCOT, and analyzed cellular neighborhoods (CNs) and spatial proximity in CODEX data.
Results: We sequenced 363,265 high-quality cells, including 65,310 CAR+ T cells. CART19 IPs were predominantly CD4+ T cells (70% total cells); CR and PD IPs did not differ in T-cell subset proportions (all P>0.05). At peak expansion, both groups showed an increase in CAR+ CD8+ effector-memory (TEM) cells from IP to peak. In CR patients, the proportion of CD8+ TEM cells was numerically higher in the CAR+ than CAR- compartment (75% vs 50%, P=0.051); no CAR-associated difference was observed in PD patients. Cell-state modeling inferred trajectories from proliferating and CD8+ naive states toward TEM within the CAR+ compartment. At peak, CAR+ CD8+ TEM cells from CR patients upregulated memory-, activation-, and TNF-associated genes (SELL, CD69, CXCR4, TNFAIP3), whereas those from PD patients upregulated IFN-associated genes (STAT1, IFIT1, IRF7). GSEA showed TNF/NF-κB enrichment in CR and type I/II IFN response enrichment in PD CAR+ CD8+ TEM cells (FDR<0.05).
Myeloid cells showed concordant response-specific pathway differences. In PD patients, DEG analysis and GSEA revealed type I/II IFN response enrichment at multiple timepoints from baseline through day 28 (ISG15, IFITM1, MX1, IRF7), whereas dendritic cells from CR patients showed peak enrichment of TNF/NF-κB signaling (FDR<0.05). These concordant associations in myeloid and CAR+ T cells prompted crosstalk analysis.
CCC analysis identified response-associated TNF family, IFN, and chemokine interactions between myeloid and CAR+ T cells. T-cell-derived TNF family interactions varied over time: TNF–TNFR1/2 was associated with PD before treatment, whereas LTA/LTB-associated signaling was associated with CR after treatment. IFNG–IFNGR1/2 and CCL4–CCR5 interactions between myeloid and CAR+/CAR- T cells were elevated in PD before and after infusion, respectively. These findings are consistent with a pre-existing, persistent IFN-associated state contributing to CAR+ T dysfunction in PD, and post-treatment TNF family signaling supporting CAR+ T function in CR.
Finally, CITE-seq surface-marker profiles in blood correlated with CODEX profiles of corresponding tumor-resident cell types (CD8 TEM: P=0.001, Myeloid: P=0.035), enabling cross-compartment comparison. Tumors contained multiple CNs with myeloid–T-cell colocalization. In PD patients, CD16+CD14+ tumor-infiltrating monocytes with a reduced immunogenic MHC class I/II-low phenotype were significantly closer to CD8+ TEM cells (P=0.008). Apheresis PBMCs from PD patients showed concordant myeloid and T-cell states, suggesting that peripheral phenotypes may reflect tumor-associated myeloid–T-cell organization.
Conclusion: Our data identify divergent TNF–NF-κB and IFN programs in CAR+ CD8+ TEM and myeloid cells from patients with durable CR versus PD. PD was associated with a persistent IFN-associated inflammatory state, whereas CR was associated with TNF family signaling post-infusion. Concordance between peripheral and spatial tumor analyses implicates myeloid–T-cell interactions in determining CART19 outcomes.