Diagnostic Performance of Host-based Gene Expression Diagnostics in Children With Extrapulmonary Tuberculosis: A Systematic Review

Background: Diagnosing extrapulmonary tuberculosis (EPTB) in children is challenging due to nonspecific presentations and poor diagnostic yield from conventional microbiologic tests. Host gene expression signatures offer a non-sputum-based diagnostic alternative. This systematic review evaluates their diagnostic performance in pediatric EPTB. Methods: We systematically reviewed host-based gene expression diagnostics for pediatric EPTB. PubMed, Embase and Cochrane Library (January 1965–May 2025) were searched for studies in children (0–18 years) with EPTB. Exclusions were adult-only studies, mixed data on pulmonary TB and EPTB without disaggregation, pulmonary TB-only studies, reviews and abstracts. Two reviewers screened data, resolving disagreements by discussion. Results: Of 830 records, 2 studies met the inclusion criteria: Pan et al. (2017) and Olbrich et al. (2024), both in low and middle-income countries, enrolling a total of 891 children under 15 years. Olbrich et al.’s 3-gene MTB-HR prototype showed 59.8% sensitivity against a strict culture-confirmed reference standard and 50.0% in isolated EPTB with a low risk of bias. Using a microbiologic, clinical and radiologic composite standard, Pan et al.’s miRNA-29a assay achieved 67.2% sensitivity, 88.5% specificity in peripheral blood mononuclear cells; 81.1% sensitivity, 90.0% specificity in cerebrospinal fluid; 84.4% sensitivity, 95.4% specificity in combined peripheral blood mononuclear cell/cerebrospinal fluid with a high risk of bias. Conclusions: Evidence for host gene expression diagnostics in pediatric EPTB is limited by few studies, small sample sizes, bias and lack of disaggregated data, with accuracy falling short of the World Health Organization targets.