Tuberculosis diagnosis in HIV-positive individuals remains one of infectious disease medicine's persistent challenges, with conventional methods often failing when immune systems are compromised. This diagnostic gap has profound implications for the 1.5 million people who die annually from TB, particularly those co-infected with HIV who face the highest mortality risk.
Ugandan researchers analyzed nasal cell gene expression in 40 HIV-positive adults, comparing 20 with confirmed tuberculosis to 20 without the disease. The nasal sampling approach identified 44 differentially expressed genes, yielding machine learning models with cross-validated area under the curve values between 0.87-0.90 for TB prediction. A streamlined four-gene signature incorporating SPIB, SHISA2, TESPA1, and CD1B genes demonstrated particular promise for clinical application. Notably, nasal and blood gene profiles showed minimal overlap, with only three shared differentially expressed genes.
This nasal biomarker approach addresses a critical clinical need where traditional sputum-based diagnostics frequently fail in immunocompromised patients. The upper respiratory tract represents the initial site of TB infection, potentially capturing disease signatures before systemic manifestation. However, the study's modest sample size and single-site design limit generalizability across diverse populations and TB strains. The work builds on emerging evidence that mucosal immunity provides distinct diagnostic windows compared to systemic immune responses. While these findings suggest nasal sampling could revolutionize TB screening in resource-limited settings, validation in larger, multicenter cohorts remains essential before clinical implementation.