Divergent Population Dynamics and Gene Signaling in Monocytes Distinctly Influence Responses to CD19-Targeted CAR T-Cells in Relapsed/Refractory B-NHL

Introduction: Emerging evidence suggests that immune dysregulation drives resistance to cellular therapies. In particular, low CAR T-cell expansion and inferior patient outcomes have been associated with an inflamed immune milieu. Here, we investigated whether myeloid cells and serum proteomics are linked to T-cell dysfunction and treatment failure.

Methods: Patients with r/r B-NHL who underwent treatment with axi-cel, brexu-cel, or tisa-cel in the third- or later-line setting were retrospectively included (n=74). Aliquots of EDTA-anticoagulated peripheral blood and serum were collected. Immune checkpoint (IC) expression (PD-1, TIM-3, and LAG-3) was assessed through flow cytometry before CAR T-cell infusion (time of leukapheresis, day –5 and day 0) as a surrogate for T-cell exhaustion. More specifically, the abundance of different co-expression profiles of ICs was compared and defined as non-exhausted (PD-1–TIM-3–LAG-3–), progenitor-exhausted (PD-1+TIM-3+LAG-3–), and terminally-exhausted (PD-1⁺TIM-3⁺LAG-3⁺). Using the Olink® Immuno-Oncology panel, 92 immune-related proteins were measured at day 0. A next-generation single-cell proteogenomics approach (BD Rhapsody^TM) was applied to analyze the transcriptome of myeloid cells before and after infusion (day –5 and day 7). Responding patients (R, complete or partial remission) were compared to non-responding patients (NR, stable or progressive disease) according to 3-month (PET-) CT scans.

Results: At baseline, NR showed lower frequencies of non-exhausted (apheresis, p = 0.025; day –5, p = 0.0236; day 0, p = 0.0496) and higher frequencies of terminally-exhausted T-cells (–5, p = 0.0053) compared to R. Interestingly, these surrogates for pre-existing T-cell dysfunction in NR were linked to an inflammatory state. Patients with low levels of non-exhausted T-cells also had higher levels of CRP (p = 0.0059), Ferritin (p = 0.0333), and LDH (p = 0.0268) compared to patients with high levels of non-exhausted T-cells before infusion.

Network analysis correlating the abundance of immune-modulatory proteins with baseline IC expression on T cells revealed numerous connections for NR but merely associations for R, suggesting an impact of the serum protein milieu on IC expression in NR. We dissected these networks in terms of the degree centrality differences for the different proteins (i.e., the variations in the number of connections among the nodes between NR and R). While we identified a higher degree centrality for co-stimulatory proteins such as IL-18, CD8A and ICOSLG in R, we noted higher degree centrality for immune-inhibitory and tumor-related proteins such as CXC3L1, TNFRSF21, VEGFA and IL-6 in NR. Accordingly, we linked a higher abundance of CD8A and a lower abundance of TNFRSF21 and IL-6 to the frequency of non-exhausted T-cells at baseline using a linear regression model.

Finally, we asked whether the myeloid compartment at baseline drives these differences in the serum protein milieu observed between NR and R. We identified four phenotypic and transcriptomic different monocytic and dendritic cell populations. Interestingly, two classical monocyte-like populations (CD14+CD16+HLA-DRlo) showed divergent population dynamics in NR but not R when assessing pre- and post-infusion time points. Notably, gene set enrichment analysis revealed an upregulation of immunosuppressive pathways, such as genes regulated by NF-KB in response to TNF, in several myeloid populations in NR but not R (classical monocyte-like 2, p < 0.0001; non-classical monocyte-like, p < 0.0001; DC-like, p = 0.0024). Surprisingly, monocytic populations in R revealed an upregulation of genes with immune stimulatory functions linked to responses to interferon-alpha or -gamma proteins (classical monocyte-like 1 and 2, p = 0.0058 and p = 0.0003; non-classical monocyte-like, p = 0.0003).

Conclusion: These data suggest that CAR-T non-responders exhibit pre-existing T-cell dysfunction resulting from a systemic inflamed proteomic and cellular environment. Notably, the profound alterations within the myeloid compartment highlight diverging immune-modulating functions in non-responding and responding B-NHL patients. These findings set the further scope for scientific investigations not only to mitigate immune dysregulation but also to target the innate immune compartment for enhanced CAR T-cell responses.