OBJECTIVES: Pediatric tuberculosis (TB) diagnosis is complicated by challenges in obtaining invasive respiratory specimens that frequently contain few Mycobacterium tuberculosis (Mtb) bacilli. We report the diagnostic performance of an Mtb antigen-derived peptide (MAP-TB) assay and its ability to monitor TB treatment response. METHODS: Study cohorts enrolled children who presented with presumptive TB at two hospitals in South Africa from 2012 to 2017 (157 children aged <
13 years) and at community-based clinics in the Dominican Republic from 2019 to 2023 (101 children aged <
18 years). Children were evaluated for TB at enrollment and six months post-enrollment and assigned confirmed, unconfirmed, or unlikely TB diagnoses using the 2015 NIH diagnostic criteria for pediatric TB. MAP-TB assay performance was evaluated using serum collected at baseline and at regular intervals post-enrollment following STARD guidelines. RESULTS: MAP-TB sensitivity for confirmed and unconfirmed TB was comparable to culture and Xpert sensitivity for confirmed TB, but MAP-TB specificity revealed age-dependence, decreasing from 98·1% to 78·4%, when including children aged <
1 year. MAP-TB values decreased by six months post-treatment initiation in children with symptom improvement. CONCLUSIONS: Serum MAP-TB results can effectively diagnose pediatric TB, including unconfirmed and extrapulmonary TB missed by current methods, and correspond to effective treatment.