Integrated Transcriptomic Analysis Reveals Candidate Biomarkers and Potential Therapeutic Targets for Tuberculosis-HIV Co-infection

Tuberculosis and HIV co-infection constitute an important global health burden, with approximately 187,000 tuberculosis-related deaths annually among people living with HIV. Despite clinical urgency, current biomarkers demonstrate limited sensitivity and specificity for detecting active tuberculosis among HIV positive individuals. This study, therefore, conducted an integrated analysis of transcriptomic data from four independent datasets encompassing 670 samples: 185 TB-HIV co-infected patients, 162 HIV mono-infected individuals, 179 tuberculosis-only patients, and 144 healthy controls from South Africa, Uganda, the United Kingdom, and India. Following rigorous quality control and batch effect correction using ComBat (sva package v3.46.0, R v4.3.0), differential expression analysis (limma) and Gene Set Enrichment Analysis (GSEA) were performed according to Broad Institute standard protocols. The analysis identified 34 robust biomarkers with an area under the receiver operating characteristic (ROC) curve of 0.947 (95% CI: 0.923 0.971), sensitivity of 88.9%, and specificity of 93.6%. GSEA revealed 47 significantly enriched biological pathways (FDR < 0.001), including interferon-gamma response (NES = 2.89), tryptophan catabolism (NES = 2.67), and IDO1 as a central metabolic checkpoint. Network topology analysis identified 12 high-confidence therapeutic targets, including CXCL10 and IDO1. Cross-validation across independent cohorts confirmed classification accuracy of 91.2%.