Exploring the relationship between immune heterogeneity characteristic genes of rheumatoid arthritis and acute myeloid leukemia

Abstract

Background: People with autoimmune diseases are prone to cancer, and there is a close relationship between rheumatoid arthritis (RA) and acute myeloid leukemia (AML). The bone marrow (BM) is affected throughout the course of RA, with a variety of hematologic involvement. Hopes are pinned on rheumatoid arthritis research to obtain BM biomarkers for AML.

Methods: Synovial transcriptome sequencing data for RA and osteoarthritis (OA), and single-cell sequencing data for RA and controls were obtained from the GEO database.Bone marrow sequencing data for AML patients and normal subjects were obtained from the UCSC Xena database. The final immune heterogeneity characteristics of RA were determined through ssGSEA analysis, gene differential expression analysis, fuzzy c-means clustering algorithm, and XGboost algorithm. Random Ferns classifiers (RFs) are used to identify new bone marrow markers for AML.

Results: SELL, PTPRC, IL7R, CCR7, and KLRB1 were able to distinguish leukemia cells from normal cells well, with AUC values higher than 0.970.

Conclusion: Genes characterizing the immune heterogeneity of RA are associated with AML, and KLRBA may be a potential target for AML treatment.

Keywords: Acute myeloid leukemia; Biomarkers; Machine learning; Therapeutic target.

Fig. 1

Results of two gene differential expression analyses (A) Immune infiltration status of all samples (B) Number of differential genes up-regulated and down-regulated

Fig. 2

Functional and pathway enrichment analysis of differentially expressed genes in RA and OA (A) Number of overlapping differentially expressed genes in RA and OA before and after immune infiltration assessment (B) Protein Interaction Network Screening for pivotal genes (C) Functional and pathway enrichment analysis of pivotal genes

Fig. 3

Identify key genes (a) Fuzzy C-means clustering results (b–e) The importance of genes in each cluster

Fig. 4

Determination and validation of gene expression and search for the optimal parameters of RFs models (A) Expression of IFNG, CXCL9, and CXCL13 in AML bone marrow (B–F) Expression of IL7R, CCR7, KLRB1, SELL, PTPRC in AML bone marrow and other tumors (G) Validation of IL7R, CCR7, KLRB1, SELL, PTPRC expression (H) Validation of expression of SELL, PTPRC, IL7R, CCR7, and KLRB1 using BioGPS database (I–M) Determine the optimal parameters for RFs classification models

Fig. 5

Diagnostic value of bone marrow biomarkers in AML (A) ROC analysis results of bone marrow biomarkers in AML (B–D) UMAP, PCA, t-SNE analysis results (E–G) Selection of the optimal parameter ‘depth’ for random fern model (H) ROC analysis results of traditional AML biomarkers

Fig. 6

Immune infiltration analysis results (A) The correlation between diagnostic genes and immune cells (B–E) Correlation between diagnostic genes and immune checkpoints

Fig. 7

GSCA online analysis and single cell sequencing analysis results (A) Drug sensitivity analysis results (B) SNV analysis results (C) Building a ceRNA network (D) Single cell sequencing analysis results in the control group (E) Single cell sequencing analysis results of RA group (F) Expression distribution of target genes in various cells of the control group (G) Expression distribution of target genes in various cells of the RA group (H) CD45 signaling pathway in the control group (I) CD45 signaling pathway in the RA group

Fig. 8

The relationship between different gene expressions and AML subtypes (A–E) Differential expression of different genes under different cytogenetic risk groupings (F) Differential expression of different genes under different CEBPA mutation types (G) Survival and prognosis analysis of patients with wild-type mutations under different gene expression patterns (H) Survival and prognosis analysis of patients with mutanted mutations in different gene expression patterns