Host plasma protein biomarkers for tuberculosis disease screening in febrile adults in Tanzania.

Abstract

Host circulating biomarkers may enhance access to tuberculosis (TB) diagnostics in low-resource settings. We sought to identify host plasma proteins that differentiate TB disease from other infectious causes of fever. This secondary analysis of a prospective cohort included outpatients ≥18 years in urban Tanzania presenting with ≤7 days of fever. Fourteen plasma proteins reflecting endothelial and immunoregulatory pathways were evaluated against a composite reference standard including sputum Xpert MTB/RIF, urine lipoarabinomannan, and/or chest x-ray. Multivariable models assessed proteins and symptoms associated with TB diagnosis. Of 507 participants, 40 (7.9%) had TB disease and 467 (92.1%) had other causes of fever. Eight proteins were significantly elevated (p<0.05) in people with TB. Regression modeling identified a four-protein biosignature (sTREM-1, CHI3L1, sTNFR-1, and CRP) with a cross-validated median AUC of 0.81 (2.5th-97.5th percentiles: 0.65-0.93), sensitivity of 80.0% (2.5th-97.5th percentiles: 41.4-100%), and specificity of 65.5% (2.5th-97.5th percentiles: 54.2-76.6%); however, discrimination was lower in people living with HIV (AUC 0.71; 95% CI 0.61-0.81). Classification and regression tree analysis yielded a simplified algorithm incorporating cough and sTREM-1, with a cross-validated AUC of 0.79 (95% CI: 0.69-0.88), sensitivity of 77.5% (95% CI: 61.6-89.2%), and specificity of 84.2% (95% CI: 80.5-87.4%); this approach may be more pragmatic for low-resource settings. This exploratory analysis identified a parsimonious biosignature and biomarker-based algorithm for TB evaluation among febrile adults in a high-burden setting. With further development, host protein-based assays may enhance TB case detection in resource-limited settings.