Introduction: Approximately 60% of Large B cell Lymphoma (LBCL) patients that receive CD19 CAR T cell therapy with axicabtagene ciloleucel (axi-cel) experience lymphoma progression (Locke et al. Lancet Oncol. 2019) and the likelihood of response to subsequent therapy is low (Spiegel, Dahiya et al. ASCO 2019). Target loss of CD19 is observed in less than a third of patients experiencing relapse. Alternative mechanisms of resistance to axi-cel are poorly understood. Lymphoma patients with elevated serum markers of systemic inflammation, such as ferritin and IL-6, have worse outcomes following axi-cel (Locke, Neelapu et al. Mol.Ther.2017; Faramand et al. ASH 2018). We hypothesized that suppressive monocytic myeloid derived suppressor cells (M-MDSCs), which are associated with worse chemotherapy outcomes in LBCL (Azzaoui et al. Blood 2016), and tumor driven inflammation may be present and responsible for decreased efficacy of axi-cel in LBCL.

Methods: LBCL patients undergoing axi-cel treatment were enrolled onto prospective sample collection protocols. Patients were stratified for analysis into ongoing responders (complete response or partial response) or relapsed (progressive disease) after a minimum of 3 months follow-up (range 3 - 15 months). M-MDSCs, defined as a Lin-, CD11b+, CD33+, CD15-, CD14+, HLA-DRlow population, were sorted from leftover apheresis material after collection for axi-cel manufacture. M-MDSC ability to suppress proliferation of autologous T cells stimulated with CD3/CD28 coated beads was measured by 3H thymidine incorporation. Circulating peripheral blood M-MDSCs, quantified by % of live cells by flow cytometry, were measured at the time of apheresis and serially after axi-cel infusion until day 30. In vitro mouse experiments utilized a CD19-CD28 CAR and cytokine-induced bone marrow MDSCs (Thevenot et al. Immunity 2014). Cytokines were measured by ELISA and cytotoxicity against CD19 bearing cell lines used xCELLigence real-time cell analysis, as we have done previously (Li et al. JCI Insight 2018).Tumor biopsies were taken within 1 month prior to infusion of axi-cel. Limited gene expression profiling of tumor microenvironment (TME) genes used the Nanostring IO360 panel (770 genes). Analysis used nSolver to identify cell types, GSEA and differential gene expression between groups.

Results: First, we demonstrated that M-MDSCs sorted from patient apheresis material suppressed the proliferation of autologous T cells (n=6). We next enumerated M-MDSCs in the peripheral blood (n = 32). M-MDSC numbers initially decreased after lymphodepleting chemotherapy but recovered to baseline levels by day +10. The level of M-MDSCs following CAR T cell therapy strongly correlated with pre-CAR T baseline levels (R = 0.871, p <0.0001), suggesting that the number of M-MDSCs present during CAR T cell expansion is dependent on factors already present before therapy began. M-MDSC levels were significantly higher in patients who subsequently relapsed, both at baseline (p= 0.01) and after axi-cel (p=0.04), as compared to patients with durable response. Mouse MDSCs were able to suppress CAR T cell IFN-gamma excretion (p<0.0001) and cytotoxicity (p<0.0001) in vitro. To evaluate the role of the TME we interrogated limited set gene expression profiling on patient (n=27) pre-axi-cel tumor biopsies. By cell type scoring, the macrophage gene score was significantly higher in patients who relapsed after CAR T therapy (p <0.001). By differential gene expression and gene set enrichment, patients who relapsed had a significantly higher expression (p <0.01) of multiple genes indicative of chronic interferon (IFN) signaling including higher levels of OAS2, OAS3, IFI6 and IFIT1, as well as the IFN-stimulated macrophage gene SIGLEC-1/CD169.

Conclusions: Systemic inflammatory myeloid cytokines, circulating M-MDSCs in the blood and chronic IFN in the TME all associate with LBCL relapse after axi-cel CAR T cell therapy. Our observations support that CAR T cells can be suppressed by baseline patient and tumor-related factors and strategies to overcome these factors should be targeted to improve patient outcomes.

MDJ and HZ contributed equally.


Jain:Kite/Gilead: Consultancy. Bachmeier:Kite/Gilead: Speakers Bureau. Chavez:Novartis: Membership on an entity's Board of Directors or advisory committees; Genentech: Speakers Bureau; Kite Pharmaceuticals, Inc.: Membership on an entity's Board of Directors or advisory committees; Janssen Pharmaceuticals, Inc.: Speakers Bureau. Shah:Jazz Pharmaceuticals: Research Funding; Incyte: Research Funding; Kite/Gilead: Honoraria; Celgene/Juno: Honoraria; Pharmacyclics: Honoraria; Adaptive Biotechnologies: Honoraria; Spectrum/Astrotech: Honoraria; Novartis: Honoraria; AstraZeneca: Honoraria. Mullinax:Iovance: Research Funding. Davila:Celgene: Research Funding; GlaxoSmithKline: Consultancy; Precision Biosciences: Consultancy; Novartis: Research Funding; Atara: Research Funding; Bellicum: Consultancy; Adaptive: Consultancy; Anixa: Consultancy. Locke:Kite: Other: Scientific Advisor; Novartis: Other: Scientific Advisor; Cellular BioMedicine Group Inc.: Consultancy.

Author notes


Asterisk with author names denotes non-ASH members.