Background/Purpose: Systemic juvenile idiopathic arthritis (SJIA) is a clinically heterogenous disease and can be complicated by macrophage activation syndrome (MAS) and lung disease (LD) thought to be driven by interferon signaling, though the contributing cell populations and distinctions between IFNγ and IFNα/β are undefined. To identify novel prognostic transcripts and potential patient subtypes, we aimed to characterize single-cell heterogeneity and patient-specific transcriptomics responses from the peripheral blood of children with SJIA, MAS and LD.

Methods: 10x Genomics single-cell RNA Sequencing (scRNA-Seq) was performed on PBMCs from 7 active SJIA and 5 inactive SJIA patients, 2 SJIA-MAS and 6 SJIA-LD patients and 5 healthy controls. Integration analyses were performed with the software Seurat 3 to identify discrete cell populations while correcting for donor and disease-level differences. To identify subsets of patients with cell-type specific signatures, we developed a new hybrid supervised/unsupervised computational pipeline in the software AltAnalyze, called cellHarmony 2.0, designed specifically for large cohort single-cell genomic studies.

Results: scRNA-Seq analysis was performed on a total of 234,128 individual cells (ranging from 6,662-12,647 cells/patient), with a mean number of 21,637 genes detected per sample. To assess cell-population level differences, we identified and annotated based on marker genes 34 discrete cell populations across all submitted samples. This indicated a consistent increase in Natural Killer (NK) cells and decrease in naïve and regulatory T cells in SJIA-LD, with the distribution of cells from inactive SJIA patients similar to that of controls. To exploit anticipated heterogeneity within this cohort, we applied our new cell-type aware patient subtype discovery algorithm cellHarmony 2.0. We computed an aggregate cell signature for all cell populations or pseudo-bulks (n=34) for each patient and their associated fold differences relative to matched control cell clusters, and performed unsupervised clustering of the pseudo-bulks to identify patient subtypes associated within one or more cell types. This analysis uncovered 11 pseudo-clusters of cell type gene expression differences, both shared and unique across the patients. Specifically, pseudo-cluster 4 was defined by IL-2 mediated signaling genes, composed of mostly NK cells from all SJIA subtypes except MAS. SJIA-MAS PBMCs were almost exclusively represented in three separate pseudo-clusters that contained genes mediating T-cell receptor activation, immune response and interferon signaling. Finally, pseudo-cluster 8 was composed of mainly monocytes/macrophages with specific upregulation of IFNα/β induced genes IFITM3, IFI6 and ISG15, only in active SJIA, SJIA-MAS and SJIA-LD.

Conclusion: Unsupervised single-cell cohort analysis provides new opportunities to uncover novel disease molecular programs and pathways in clinically heterogenous patient groups. Here, we found active SJIA, SJIA-MAS and SJIA-LD PBMCs have distinct monocytic responses characterized by upregulation of interferon-induced genes, highlighting the role for both IFNγ and IFNα/β in driving disease pathogenesis.

« Back to ACR Convergence 2021

ACR Meeting Abstracts - https://acrabstracts.org/abstract/single-cell-genomics-reveals-a-shared-monocyte-interferon-program-in-a-subset-of-patients-with-systemic-juvenile-idiopathic-arthritis-macrophage-activation-syndrome-and-lung-disease/