Effect of interferon induced transmembrane protein 1 (IFITM1) upregulation to cytokine release syndrome in CAR-T-treated B-cell acute lymphoblastic leukemia

To the Editor: Chimeric antigen receptor (CAR)-T cells targeting CD19 have transformed the treatment of relapsed or refractory B-cell acute lymphoblastic leukemia (R/R B-ALL) therapy.[1] However, cytokine release syndrome (CRS) represents a serious and potentially life-threatening complication, marked by excessive proinflammatory cytokine release, poses a severe, potentially fatal complication, manifesting as fever, nausea, fatigue, and, in severe cases, multi-organ failure.[2,3] Advances in single-cell transcriptomics have enabled high-resolution analysis of complex tissues, uncovering the cellular dynamics underlying CRS. For example, single-cell protein secretion profiling of pre-infusion CD19 CAR-T products in non-Hodgkin lymphoma patients revealed differential cytokine and chemokine expression, shedding light on individual CRS variability.[4] Despite these breakthroughs, the role of bone marrow mononuclear cells (BMMCs) at the single-cell level in modulating CRS outcomes has remained insufficiently explored. This study aimed to explore the heterogeneity of BMMCs and their influence on inter-patient variability in CRS severity following anti-CD19 CAR-T therapy, which may influence the development of targeted therapeutic strategies, advancing the safety and efficacy of CAR-T therapies. This study was approved by the Ethics Committee of the Union Hospital, affiliated with Huazhong University of Science and Technology, Wuhan, China (No. [2019]008) and was conducted in compliance with the Declaration of Helsinki. Written informed consent was obtained from all participants. The study has been registered on http://ClinicalTrials.gov (No. NCT04008251). This study enrolled 32 subjects with B-ALL, with baseline covariates detailed in Supplementary Table 1, https://links.lww.com/CM9/C429. The inclusion and exclusion criteria for patient selection and detailed methodology are available in the Supplementary Method, https://links.lww.com/CM9/C429. Data collection on safety and efficacy was conducted from October 2020 at Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, and concluded on May 5, 2023. CRS was observed in 18 participants, including 7 cases of grade-3 or higher severity, as presented in Supplementary Table 1, https://links.lww.com/CM9/C429. For the 7 cases, treatment interventions included tocilizumab administration for 4 subjects and corticosteroids for 3 subjects. The median interval from CAR-T-cell infusion to CRS onset was 7 days (range: 3–14 days), with a median CRS duration of 5 days (range: 1–19 days) among the 18 patients. Single-cell RNA-Sequencing (scRNA-seq) samples were collected from 4 patients at 3 different time points: baseline BMMCs (BMMC-1), infusion products (IPs), and BMMCs at peak CRS (BMMC-2). Among the 4 subjects, 2 experienced grade-0/1 CRS, while 2 had grade-3 or higher CRS, as illustrated in Supplementary Figure 1, https://links.lww.com/CM9/C429. The scRNA-seq approach employed in this study is outlined in Supplementary Figure 2A, https://links.lww.com/CM9/C429. A total of 120,780 residual cells from BMMC-1, IPs, and BMMC-2 of the 4 patients were analyzed using 5′-transcriptomic single cell RNA sequencing. Due to the independent processing of each sample for library preparation, a formal assessment of batch effects was not feasible, as each sample was considered an individual batch. However, cells from all samples were distributed across multiple clusters, with no cluster restricted to a single sample, minimizing concerns about batch effects [Supplementary Figure 2B, https://links.lww.com/CM9/C429]. Following unbiased clustering of all BMMCs and IPs, 21 transcriptionally distinct cell clusters were identified. Cluster annotations were determined using previously established marker genes.[5] These clusters were further grouped into five primary categories: IPs, hematopoietic stem and progenitor cells (HSPCs), natural killer/T cell (NK/T) cells (including T cells, NK cells, and NK-T cells), myeloid cells (comprising dendritic cells, monocytic lineages, and granulocytic lineages), and B-ALL cells [Supplementary Figure 2C, https://links.lww.com/CM9/C429]. CAR expression was distributed uniformly across the infusion product (IP) cluster [Supplementary Figure 2D, https://links.lww.com/CM9/C429] without significant gene expression differences when compared to non-transduced T cells [Supplementary Figure 3A, https://links.lww.com/CM9/C429]. As a result, CAR-positive and CAR-negative cells were not analyzed separately in subsequent evaluations, consistent with prior findings.[5] To further investigate the contributions of various T cell subtypes, IPs were classified into CD4+CD8– T cells, CD4–CD8+ T cells, double-positive T (DPT) cells, and double-negative T (DNT) cells [Supplementary Figure 2D, https://links.lww.com/CM9/C429].[5] HSPCs, NK/T cells, and myeloid cells were categorized using well-characterized gene signatures, including lineage-specific transcription factors, effector genes, and surface markers [Supplementary Figure 2E, https://links.lww.com/CM9/C429]. B-ALL clusters were identified based on clinical leukemia immunophenotyping using surface markers from bone marrow samples, assessed via flow cytometry [Supplementary Figure 2F, https://links.lww.com/CM9/C429]. Comparative gene set enrichment analysis (GSEA) of functional features across these five cell types further validated the cell annotation results [Supplementary Figure 3, https://links.lww.com/CM9/C429]. Upregulated genes associated with high-grade CRS were identified through comprehensive analyses using Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. BMMCs from patients with severe CRS demonstrated significant enrichment in genes implicated in diverse cytokine-associated signaling pathways [Figure 1A]. Notably, in B-ALL cells from the high disease burden (DB) group, elevated gene expression was evident across nearly all analyzed pathways. In contrast, NK/T and myeloid cells from the low DB group displayed upregulation of genes involved in cytokine biosynthesis, metabolism, secretion, and the positive regulation of cytokine production. Enrichment in IP subsets was limited to a selected number of pathways, predominantly those regulating cytokine production.Figure 1: Transcriptomic regulation of cytokines associated with CRS in the bone marrow immune microenvironment. (A) Heatmap depicting cytokine-related signaling pathways that were significantly upregulated for each cell type. (B) Heatmap illustrating detailed cytokine-related enriched pathways by comparing high-grade CRS with low-grade CRS patients. (C) Protein-protein interaction network reflecting the molecules implicated in B-ALL and HSPC cells in the context of high disease burden and NK/T and myeloid cells in the context of low disease burden. (D) Volcano plots describing the expression differentiation between the high-grade CRS (n = 5, left, blue) and low-grade CRS (n = 8, right, red) patients from the bulk RNA sequencing group at BMMC-1. B-ALL: B-cell acute lymphoblastic leukemia; BMMC: Bone marrow mononuclear cell; CRS: Cytokine release syndrome; DB: Disease burden; DPT: Double-positive T cells; DNT: Double-negative T cells; GO/KEGG: Gene Ontology/ Kyoto Encyclopedia of Genes and Genomes; HSPC: Hematopoietic stem and progenitor cell; IFITM1: Interferon-Induced Transmembrane Protein 1; NK/T: Natural killer/T cells; t-SNE: t-distributed stochastic neighbor embedding plot.To delineate specific cytokines contributing to severe CRS, 44 distinct cytokine-associated signaling pathways were identified through advanced pathway analysis [Figure 1B]. Pathways associated with interferon (IFN) production and response represented the most enriched group, comprising 11 individual pathways. Following IFN-related pathways, interleukin (IL)-1 were enriched in seven distinct signaling processes. Other cytokines, including IL-6, IL-8, and IL-15, exhibited enrichment in one to four pathways. Differentially expressed genes (DEGs) demonstrated distinct distribution patterns between the high and low DB groups. Specifically, B-ALL cells and HSPCs in the high DB group exhibited significant enrichment for IFN- and IL-1-related pathways. Conversely, NK/T and myeloid cells in the low DB group were predominantly enriched for pathways associated with IFN, IL-1, IL-8, and IL-12, reflecting a broader and more heterogeneous cytokine signaling landscape influenced by disease burden. DEGs related to cytokine signaling in B-ALL and HSPCs from the high DB group, as well as in NK/T and myeloid cells from the low DB group, were used to construct protein-protein interaction (PPI) networks. Hub genes were identified based on stringent criteria, including high interaction centrality (top 50%), substantial fold changes (log2 fold change >2), and elevated expression levels (average expression >0.5). Except for two genes encoding tumor necrosis factor (TNF) superfamily receptors, all hub genes were associated with IFN-related signaling pathways across all cell types. Among these, interferon-inducible transmembrane protein 1 (IFITM1) emerged as a pivotal hub gene within each analyzed dataset [Figure 1C]. To validate these findings, bulk RNA sequencing was conducted on baseline BMMC samples from 28 patients, comprising 23 patients with low-grade CRS and 5 patients with high-grade CRS. Differential expression analysis revealed distinct transcriptomic profiles between the two cohorts. IFITM1 displayed a higher mean relative expression in high-grade CRS patients than in low-grade CRS patients (0.52 vs. 0.11, P <0.001; Figure 1D) suggesting its critical role in modulating CRS severity. In vitro experiments demonstrated that IFITM1 knockdown significantly reduced IL-6 expression in Nalm-6 cells or THP1 cells when co-cultured with CAR-T cells (both P <0.001). [Supplementary Figure 4A and B, https://links.lww.com/CM9/C429]. These results suggested the cell-type-specific impact of IFITM1 knockdown on IL-6 expression. Tumor burden plays a pivotal role in determining the severity of CRS. Interestingly, patient 18, despite presenting with a high proportion of bone marrow blasts both pre- and post-CAR-T infusion, experienced only grade-1 CRS on day 14. This observation suggested that the phenotypic characteristics of tumor cells might also contribute to CRS pathogenesis. To investigate this, B-ALL cells were classified into 4 transcriptional clusters (clusters 1, 3, 5, and 10) based on predefined annotations [Supplementary Figure 5A, https://links.lww.com/CM9/C429]. Each cluster was analyzed at BMMC-1 and BMMC-2 to assess their respective contributions to CRS in patient 18 [Supplementary Figure 5B, https://links.lww.com/CM9/C429]. Notably, post-infusion analysis revealed an increase in cell counts within cluster 1, whereas clusters 3, 5, and 10 demonstrated a marked decline. DEGs in clusters 3, 5, and 10 were significantly enriched in apoptosis-related pathways (adjusted P<0.001) and TNF signaling pathways (adjusted P = 0.007), involving genes such as early growth response 1 (EGR1), TNF, and growth arrest and DNA damage inducible beta (GADD45B) [Supplementary Figure 5C, https://links.lww.com/CM9/C429]. In contrast, DEGs in cluster 1 were predominantly enriched in IFN-related pathways (adjusted P <0.001), with statistically significant upregulation of interferon alpha inducible protein 6 (IFI6), MX dynamin like GTPase 1 (MX1), and interferon stimulated exonuclease gene 20 (ISG20) (all P <0.001). At baseline, tumor cells with high expression of IFN-related signaling genes were extremely rare in the CAR-T infusion product. However, as these cells expanded in the bone marrow, CAR-T cells were nearly depleted, correlating with the patient’s mild CRS presentation. CRS remains a significant challenge in CAR-T therapy, and the current CRS management with IL-6 monoclonal antibodies and corticosteroids has limitations, including risks of exacerbating CRS or impairing CAR-T efficacy. Serum biomarkers such as IFN-γ have been identified as targets of the severity of cytokine storms induced by CAR T-cell therapy. The early upregulation of IFN-induced genes such as IFITM1 suggests a hyperinflammatory state. The signaling pathways involving IFN-α/β/γ, IL-1, IL-6, IL-8, IL-12, IL-18, and the TNF superfamily reciprocally enhance cytokine production, creating a positively regulated network. In such an inflammatory environment, macrophages are massively activated and transformed into the M1 type, thereby promoting the secretion of IFN-γ, leading to uncontrolled cytokine secretion. Janus kinase (JAK)1 and JAK2 inhibitors, which are approved for various autoimmune and neoplastic conditions, can block signaling pathways downstream of IFN-γ and type I IFNs, among other cytokines. A research discussed the use of neutralizing IFN-γ to enhance the safety of CAR-T cell therapy for malignant B-cell tumors. The results indicated that neutralizing IFN-γ might be a strategy to reduce immune-related side effects without sacrificing therapeutic efficacy, providing rationale for combining emapalumab with CAR-T cell therapy in clinical settings.[6] The IFITM1 gene, acting as a key molecule of the IFN signaling pathway, can serve as an early predictive biomarker for the occurrence of severe CRS and provide guidance for timely intervention. Early evaluation of CAR-T infusion outcomes and bone marrow immune microenvironment is critical. Single-cell gene regulation insights provide a foundation for identifying predictive biomarkers and personalized intervention strategies, advancing precision in CRS management. Funding This work was supported by grants from the National Natural Science Foundation of China (Nos. 82330005, 82400212, and 82200144), and the open fund of Hubei Key Laboratory of Biological Targeted Therapy Research (No. 202302). Acknowledgements We particularly thank Wuhan Si’an Medical Technology Co., Ltd. for the anti-CD19 CAR-T cells. Conflicts of interest None.