# Integrated multi-omics profiling reveals immune-related biomarkers and regulatory networks for early prediction of tuberculosis in type 2 diabetes mellitus

## TL;DR

This study identifies immune-related biomarkers and regulatory networks for early detection of tuberculosis in type 2 diabetes patients using multi-omics data.

## Contribution

The study introduces a validated multi-omics framework and actionable biomarkers for early prediction of tuberculosis in type 2 diabetes mellitus.

## Key findings

- Thirteen immune-related biomarkers were identified, including mRNAs, miRNAs, lncRNAs, and proteins.
- Eleven models demonstrated high diagnostic efficacy with AUC values ranging from 0.93 to 0.99.
- Key regulatory axes like IRF1/IFN-γ and ceRNA circuitry were elucidated for T2DM-TB.

## Abstract

Type 2 diabetes mellitus (T2DM) significantly elevates the risk of tuberculosis (TB); however, early detection in T2DM patients is still insufficient. This study aimed to identify immune-based early-warning biomarkers, develop robust prognostic models, and elucidate the immune-metabolic circuitry underlying the comorbidity of type 2 diabetes and tuberculosis (T2DM-TB).

A prospective cohort study (n = 198; HC 71, T2DM 67, T2DM-TB 60) was conducted, involving whole-transcriptome and plasma-proteome profiling. Differential expression analysis, weighted gene co-expression network analysis (WGCNA), and mining of the ImmPort database facilitated the extraction of immune-relevant genes. Protein-protein interaction (PPI) and competing endogenous RNA (ceRNA) networks were utilized to delineate core regulators. Eleven logistic regression models were developed based on 13 cross-platform biomarkers. The robustness of these models was evaluated through 5-fold cross-validation, and feature selection was optimized using least absolute shrinkage and selection operator (LASSO) regression. External validation was performed using GEO datasets (GSE181143, GSE114192) and reverse transcription quantitative polymerase chain reaction (RT-qPCR). Functional annotation and xCell immune-infiltration analyses were employed to characterize microenvironmental shifts, while dual-luciferase assays confirmed ceRNA interactions.

Thirteen immune-related biomarkers were identified, comprising 4 mRNAs (IRF1, FPR1, LILRB3, SECTM1), 2 microRNAs (miRNAs) (hsa-miR-4726-5p, novel-miR-109), 3 long non-coding RNAs (lncRNAs) (MSTRG.128052.1, MSTRG.4908.1, MSTRG.37670.90), and 4 proteins (IFN-γ, IL-6, CXCL10, CXCL6). Eleven models demonstrated high diagnostic efficacy, with area under the curve (AUC) values ranging from 0.93 to 0.99, and exhibited stable performance in 5-fold cross-validation, yielding AUC values between 0.77 and 0.95. LASSO-derived concise biomarker subsets overlapped with primary model features, thereby confirming robust discriminative stability. Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses underscored the significance of immune response, inflammation, and metabolic regulation, highlighting key pathways such as Toll-like receptors, NF-κB, and JAK-STAT. Immune infiltration analysis revealed a “pro-inflammatory-suppressive-reconstructive” imbalance characterized by overactivated innate immunity, including M1/M2 macrophages and NKT cells, alongside compromised adaptive immunity, evidenced by reduced CD4⁺/CD8⁺ T cells and B cells. Additionally, ceRNA networks and dual-luciferase assays confirmed that novel-miR-109 inhibits the translation of FPR1, LILRB3, and MSTRG.4908.1, while hsa-miR-4726-5p targets the 3’ UTR of SECTM1.

This study establishes a validated multi-omics framework for the early detection of T2DM-TB, elucidates key regulatory axes (IRF1/IFN-γ, ceRNA circuitry, CXCL10/CXCL6), and provides actionable biomarkers and high-performance models for precision intervention in T2DM-TB management.