Abstract
INTRODUCTION: Colonoscopy remains the gold standard for diagnosing inflammatory bowel disease (IBD) even though it is an invasive and costly procedure. To enable non-invasive diagnosis, we aimed to identify blood-based transcriptomic biomarkers that specifically distinguish IBD from healthy and inflammatory controls. METHODS: Public microarray and RNA-seq datasets from whole blood of IBD, rheumatoid arthritis (RA), and control subjects were analyzed. RA was included as a positive control for systemic inflammation to filter out non-IBD-specific gene signatures. Differentially expressed genes (DEGs) were identified, followed by immune cell deconvolution (CIBERSORTx), pathway and network analysis, and diagnostic model construction using LASSO and SVM. A real-life cohort (36 IBD patients, 30 controls) was recruited for qRT-PCR validation. RESULTS: IBD blood transcriptomes exhibited altered immune profiles, including increased M0 macrophages, Tregs, and CD4 naïve T cells, and decreased memory B and activated NK cells. After excluding RA-overlapping DEGs, 25 IBD-specific DEGs with |log2FC| > 0.5 were prioritized. LASSO and SVM identified a three-mRNA panel-IL4R, EIF5A, and SLC9A8-which achieved 84% diagnostic accuracy in the discovery cohort and 99% accuracy in the real-life cohort. Network analysis highlighted NDUFB2 as a key downregulated hub gene linked to mitochondrial complex I dysfunction and oxidative phosphorylation disruption. Elevated oxidative stress in IBD was confirmed by increased Total Oxidant Status (TOS) levels in patient plasma. DISCUSSION: Our findings support the use of peripheral blood transcriptomics for IBD diagnosis and demonstrate that a focused three-gene panel can achieve high diagnostic accuracy. The inclusion of RA as an inflammatory control enabled the identification of IBD-specific markers, minimizing confounding from general immune activation. These results provide a practical foundation for developing non-invasive diagnostic tools for clinical use.