Weekly Digests
‹ Back to June

Playing the blame game with cytokine release syndrome

June 27, 2018

CAR T cell therapy has proven to be an effective treatment for B cell malignancies, but treatment is often complicated by cytokine release syndrome (CRS), a severe side effect (particularly in patients with high tumor burden) that causes fever, hypotension, respiratory insufficiency, and potentially death due to spiking levels of serum cytokines. Moreover, delayed, potentially lethal neurotoxicity can follow disease clearance. Severe CRS can usually be alleviated with IL-6-blocking antibodies, but the exact mechanism underlying CRS is not fully understood. Both Giavridis et al. and Norelli et al. investigated the source of the cytokines that cause CRS, and explored possible alternative interventions to prevent or treat CRS; their papers appeared back to back in Nature Medicine.

Because most mouse models do not recapitulate CRS, Giavridis et al. and Norelli et al. each developed unique mouse models. Giavridis et al. developed a model in which CD19-targeted human CAR T cells were used to treat established intraperitoneal Raji tumors in SCID-beige mice. In this model, potentially lethal CRS symptoms developed 2-3 days after infusion of CAR T cells, a cytokine profile similar to that reported in clinical studies emerged, and symptoms were responsive to IL-6 blockade.

Due to the xenogeneic nature of this mouse model, Giavridis et al. were able to easily distinguish between cytokines derived from the human CAR T cell product and those derived from the murine host immune system. While CAR T cells were found to produce cytokines like IFNγ and GM-CSF, endogenous cells were primarily responsible for the production of mouse IL-6 (mIL-6). The researchers confirmed high levels of myeloid cells (known to produce mIL-6), and identified macrophages as the main source of mIL-6.

Interestingly, mIL-6 production was only upregulated in myeloid cells that had colocalized with tumors and CAR T cells, indicating a possible interaction at the tumor site. To investigate this, the team engineered CAR T cells that expressed murine CD40L and were therefore capable of interacting with the murine CD40 receptor on host myeloid cells. These CAR T cells demonstrated enhanced symptoms of CRS even though they recruited a similar number of myeloid cells, indicating that CAR T cells interact with host myeloid cells at the tumor site, impacting CRS.

Next, the team investigated the role of inducible nitric oxide synthase (iNOS), an enzyme produced by activated macrophages, and found that only tumor-associated macrophages produced iNOS in their model. iNOS produces nitric oxide (NO) which can cause vasodilation and hypotension — symptoms associated with CRS. Symptoms of CRS could be alleviated by inhibitors of iNOS or by blockade of either IL-1 or IL-6, which are known inducers of iNOS. Lack of a synergistic effect when blocking both IL-1 and IL-6 indicated that the two cytokines utilized a common pathway.

In an effort to alleviate CRS without the need for exogenous intervention, Giavridis et al. engineered a CAR T cell that constitutively produced the IL-1 receptor antagonist (IL-1Ra). This new construct protected against CRS-associated mortality without affecting serum levels of CAR T cell-derived cytokines or impacting antitumor efficacy.

Although the model developed by Giavridis et al. effectively recapitulated CRS, it was not used to investigate the delayed lethal neurotoxicity that can appear in patients who had previously experienced CRS. In the humanized mouse model developed by Norelli et al., both of these effects were modeled. At a newborn stage, triple transgenic NSG (SGM3) mice expressing IL-3, human stem cell factor, and GM-CSF were sublethally irradiated and subsequently transplanted with human hematopoietic stem and progenitor cells (HSPCs); this strategy supported reconstitution of a human immune system and successful thymic education of human T cells. T cells from these mice were used to engineer either CD19- or CD44-targeting CAR T cells (with different endodomain variants), which were transplanted into secondary humanized SGM3 mice. The target antigen and endodomain selected for CAR T cell design both appeared to affect antitumor efficacy and CRS severity. In SGM3 mice that had been engrafted with both human HSPCs and ALL-CM leukemia cells, CRS severity and mortality following CAR T cell therapy correlated with tumor burden. Moreover, in some survivors, a sudden, sometimes lethal neurotoxicity was observed, recapitulating the late onset of neurotoxicity observed in human CAR T cell patients.

In this model, Norelli et al. noticed an accumulation of monocytes coincident with leukemia progression, and were able to determine that CRS mortality correlated with monocyte counts. Depletion of monocytes before CAR T cell infusion prevented CRS and mortality, though it had a negative effect on in vivo CAR T cell expansion and on the kinetics of leukemia clearance, indicating a possible adjuvant role for monocytes.

Like Giavridis et al., Norelli et al. found that host immune cells, not CAR T cells, were primarily responsible for producing IL-1 and IL-6, and other cytokines implicated in CRS, though in their studies, human monocytes emerged as the primary source, rather than macrophages. Interestingly, a time course study showed that IL-1 release preceded IL-6 release by about 24 hours. As expected, blocking IL-6 prevented CRS, but was unable to prevent delayed lethal neurotoxicity. Prophylactic administration of an IL-1 antagonist, however, was able to prevent both CRS and neurotoxicity without affecting the antitumor efficacy of CAR T cells. In the therapeutic setting, administration of either IL-6 blockade or IL-1 antagonist following onset of fever decreased CRS mortality, though the IL-1 antagonist had borderline statistical significance; still, only the IL-1 antagonist protected against lethal neurotoxicity and extended leukemia-free survival.

Overall, both groups showed that host immune cells, not CAR T cells, are responsible for producing the cytokines that cause CRS, both identify IL-1 as a possible alternative target to IL-6, and both present possible intervention strategies to mitigate the potentially deadly side effects of CAR T cell therapy. Finding strategies to prevent or abate CRS and neurotoxicity are critical to improving and expanding CAR T cell therapy for the treatment of cancer.

by Lauren Hitchings


Giavridis T., van der Stegen S.J.C., Eyquem J., Hamieh M., Piersigilli A., Sadelain M. CAR T cell-induced cytokine release syndrome is mediated by macrophages and abated by IL-1 blockade. Nat Med. 2018 Jun 24.

Norelli M., Camisa B., Barbiera G., Falcone L., Purevdorj A., Genua M., Sanvito F., Ponzoni M., Doglioni C., Cristofori P., Traversari C., Bordignon C., Ciceri F., Ostuni R., Bonini C., Casucci M., Bondanza A. Monocyte-derived IL-1 and IL-6 are differentially required for cytokine-release syndrome and neurotoxicity due toCAR T cells. Nat Med. 2018 Jun 24.

In the Spotlight...

Combined Rho-kinase inhibition and immunogenic cell death triggers and propagates immunity against cancer

Nam et al. show that blockade of Rho-associated kinases (ROCKs) using ROCK inhibitor Y27632 increased phagocytosis of tumor cells in vitro and in mice with syngeneic tumors. In vivo, ROCK inhibition led to increased migratory CD103+ DC-mediated CD8+ T cell priming, IFNγ production, and tumor growth suppression. Combination of doxorubicin and Y27632 increased infiltration of CD8+ T cells into the tumor and further suppressed tumor growth, leading to increased survival, complete tumor regression, and long term memory in some mice bearing syngeneic or autochthonous tumors.

PD-1 Blockade Unleashes Effector Potential of Both High- and Low-Affinity Tumor-Infiltrating T Cells

Using low and high avidity OVA-specific transgenic T cells, bilateral implantation of tumors bearing either low- or high-affinity epitopes, and immunization with either antigen, Martinez-Usatorre et al. evaluated the role of TCR avidity in immune response and tumor control, reflecting neoantigens (high presumed avidity) and tumor-associated antigens (TAAs; low presumed avidity). Low-avidity T cells retained significant functionality in vivo and appeared less exhausted, but expanded poorly upon stimulation. Enhanced stimulation via vaccine and addition of anti-PD-1 demonstrated the in vivo tumor control potential of TAA-specific T cells.

Mass cytometry of Hodgkin lymphoma reveals a CD4+ exhausted T-effector and T-regulatory cell rich microenvironment

Using a customized time-of-flight mass cytometry (CyTOF) marker panel, Cader et al. confirmed that in classical Hodgkin lymphoma, the tumor microenvironment contains tumor cells with downregulated expression of β2M and MHC class I molecules, and revealed an enrichment of CD4+ T cells, which consist mostly of differentiated, Th1-polarized PD-1+ T effector cells and PD-1- Tregs. The differential PD-1 expression points to dual immunosuppression mechanisms – exhausted T effectors and activated Tregs.

Enhancing the potency and specificity of engineered T cells for cancer treatment

Sukumaran et al. created “SmarT” CAR T cells that target pancreatic cancer and not normal tissue by requiring independent signaling via three co-occurring events: recognition of a target antigen (prostate stem cell antigen) and presence of two immunosuppressive cytokines (TGFβ and IL4). In vivo, SmarT CAR T cells expanded only in the tumor, eradicated the tumor in a manner that required both CD4+ and CD8+ CARs, contracted shortly after tumor eradication, and persisted for recall response.

T-cell responses to TP53 "hotspot" mutations and unique neoantigens expressed by human ovarian cancers

Deniger et al. analyzed resected ovarian cancer metastases in seven women to identify the neoantigen specificity of TILs. Eight mutations generated neoantigen-specific T cells in five women, including a single mutation which generated three epitopes that could each bind a separate HLA. Three CD8+ and nine CD4+ specificities were observed. CD4+ T cells recognizing either of two recurrent TP53 mutations (p53Y220C or p53G245S) were identified in separate patients. The identification of mutation-reactive T cells infiltrating ovarian cancer metastases opens possibilities for TCR-based adoptive cell transfer.

Disruption of TET2 promotes the therapeutic efficacy of CD19-targeted T cells

Analyzing a patient with advanced relapsed/refractory chronic lymphocytic leukemia who achieved complete remission following CD19-targeted CAR T cell therapy, Fraietta et al. found that 94% of CAR T cells at the peak of expansion had originated from a single clone in which the CAR transgene had disrupted the TET2 gene in one allele, while the other allele contained a hypomorphic (partial loss of function) mutation. A majority of these cells exhibited an epigenetically-programmed antitumor memory phenotype. Experimental knockdown of TET2 enhanced the capacity of T cells to proliferate and sustain cytokine production upon restimulation.

Everything New this Week In...