Monocyte-driven inflammation is emerging as a marrow toxicity pathway after CAR T-cell therapy
Immune effector cell-associated hematotoxicity, or ICAHT, has become one of the defining complications of CAR T-cell therapy beyond cytokine release syndrome and neurotoxicity. The review frames ICAHT as more than a simple carryover from lymphodepleting chemotherapy. Neutropenia, anemia, and thrombocytopenia can persist, recur, or appear late, and the clinical burden extends to infections, transfusion dependence, delayed treatment, and longer hospitalization.
A multicenter cohort of 549 recipients of CD19- or BCMA-directed CAR T cells found grade ≥3 early ICAHT in 28% of patients with mantle cell lymphoma, 23% with large B-cell lymphoma, and 15% with multiple myeloma. Severe ICAHT tracked with more infections and greater transfusion need. The hematopoietic injury is not uniform, and the pattern of recovery matters as much as the nadir.
Why marrow recovery fails after the acute CAR T-cell surge
The central argument is biological rather than descriptive. CAR T-cell expansion creates a systemic inflammatory wave that can strike already fragile marrow reserves after prior cytotoxic therapy, baseline cytopenias, tumor infiltration, or infection. Lymphodepleting chemotherapy clearly contributes, but it does not explain the full duration or heterogeneity of cytopenias seen after infusion.
Inflammatory stress can push quiescent hematopoietic stem cells away from steady-state renewal and toward exhaustion, aberrant proliferation, lineage skewing, or cell death. Stromal and endothelial support within the marrow niche may also be injured, leaving hematopoietic stem and progenitor cells with fewer signals to recover. In this setting, acute inflammation can evolve into a prolonged suppressive state.
Monocytes and macrophages sit at the center of the inflammatory loop
Monocytes and macrophages are not passive bystanders in this process. They bridge innate immune sensing, cytokine release, tissue repair, and hematopoietic regulation. Under homeostatic conditions, marrow macrophages support stem-cell retention and erythroid maturation. Under inflammatory pressure, those same cells can be redirected toward cytokine amplification and disordered marrow function.
CAR T cells release IFN-γ, TNF-α, and granulocyte-macrophage colony-stimulating factor after antigen recognition, and those signals engage host myeloid cells. In coculture systems, CAR T-cell-derived GM-CSF induced monocyte production of IL-6 and CCL2 without direct cell contact, and neutralizing GM-CSF reduced the response. In xenograft models, GM-CSF blockade attenuated CRS and neuroinflammation while preserving antitumor activity. The pathway is important because it links T-cell activation to a host-driven myeloid cascade that can outlast the initial tumor-killing event.
Tumor-cell pyroptosis adds another layer. CAR T-cell killing can trigger caspase-3 activity in target cells, cleavage of gasdermin E, and inflammatory pyroptosis. Factors released from pyroptotic cells activate caspase-1 and gasdermin D in macrophages, which intensifies cytokine release. In mouse models, target-cell GSDME deletion, macrophage depletion, or caspase-1 inhibition prevented CRS. In patients, GSDME expression and lactate dehydrogenase levels correlated with CRS severity. One abrupt takeaway follows from those data: the tumor cell itself can become a cytokine source after it is killed.
Acute CRS biology and persistent hematotoxicity are related, but not identical
The review draws a careful boundary between the mechanism of acute CRS and the less defined biology of persistent ICAHT. Monocyte and macrophage activation is clearly causal in acute systemic toxicity in preclinical models, where human monocytes supply IL-1 and IL-6 and recipient macrophages govern CRS severity through IL-1, IL-6, and nitric oxide. Those experiments establish myeloid cells as key amplifiers of early inflammatory toxicity.
Whether the same cells drive prolonged marrow suppression after CRS resolves is still unresolved. Persistent cytopenias may reflect continued marrow inflammation, altered niche repair, ongoing endothelial dysfunction, or a delayed transition from inflammatory activation to tissue restoration. The review treats that distinction as clinically important because the mechanism likely changes over time, and the biomarker strategy has to change with it.
Translational signals from cytokines, proteomics, and risk scoring
CAR-HEMATOTOX is one practical risk framework discussed in the review. The pretreatment score integrates ANC, hemoglobin, platelets, C-reactive protein, and ferritin, with scores of 0–1 defining a lower-risk group and scores ≥2 indicating higher risk. Those inputs do not prove causation, but they do capture the intersection of compromised marrow reserve and baseline inflammation before lymphodepletion begins.
Other biomarker work points toward macrophage activation, endothelial dysfunction, and T-cell suppression in patients with an aplastic ICAHT phenotype. Longitudinal transcriptomic and serum cytokine analyses in 26 cilta-cel-treated patients showed a dynamic multicellular network: neutrophil activation preceded overt CRS, while T cells and monocytes/macrophages dominated cytokine production during established CRS. JAK/STAT pathway activation appeared before the cytokine surge. In parallel, functional work on the BCAP-RSK-eIF4B pathway showed that myeloid cells can become hypersecretory through enhanced translation, not just altered transcription. That matters because routine RNA profiling may miss a meaningful part of the inflammatory state.
Hematopoietic suppression likely reflects several overlapping marrow insults
The most plausible model is layered injury. Prior therapy narrows reserve, CAR T-cell expansion triggers a cytokine surge, and activated monocytes and macrophages extend inflammation into the marrow microenvironment. In that setting, HSCs may lose quiescence, myeloid and erythroid production may skew, and supportive niche function may falter. The review also points to IL-18 as a candidate amplifier because it can reinforce IFN-γ production and macrophage activation, although the supporting data remain limited and need larger validation.
This is also where clinical translation becomes more difficult. A marker that predicts CRS does not automatically predict prolonged cytopenia, and a therapy that suppresses inflammation early may not help if the dominant problem later is failed marrow repair. Longitudinal sampling of peripheral blood and bone marrow, multi-omics profiling, and functional validation are needed to separate acute inflammatory amplification from sustained hematopoietic suppression.
The therapeutic question is no longer whether myeloid cells matter
The remaining challenge is how to intervene without blunting CAR T-cell efficacy. GM-CSF blockade, caspase-1 inhibition, catecholamine-synthesis interference, and RSK inhibition all appear in the mechanistic discussion as possible ways to dampen myeloid amplification. Each strategy is attractive for a different reason: some target upstream cytokine release, others interrupt downstream signaling or translation. None of them yet define standard practice for ICAHT.
The review closes with a concrete translational agenda rather than a definitive treatment algorithm. Serial marrow studies, cellular phenotyping, and biomarker-driven trials will be needed to determine whether monocytes and macrophages merely accompany hematotoxicity or actively sustain it. Until those studies are done, the field has a coherent mechanism and an incomplete therapeutic map.