Blood transcriptomic signatures in children with tuberculosis in a treatment-shortening trial

<title>Abstract</title> <p>The SHINE trial demonstrated that 4 months of treatment was non-inferior to 6 months for children with non-severe tuberculosis drug-susceptible (DS-TB) as assessed on chest radiograph. However, the optimal duration likely varies between individuals. We evaluated blood transcriptomic signatures as non-sputum biomarkers of treatment response and explored their potential to inform individualised treatment duration in children. This sub-study included children (&lt;16 years) with non-severe TB enrolled at a trial site in Cape Town, South Africa. Participants were classified as confirmed, unconfirmed, or unlikely TB by a blinded endpoint review committee. Whole blood was collected in PAXgene tubes at enrolment (baseline), week 2, week 8, and at the End of Treatment (EoT; week 16 or 24). We measured 88 gene transcripts by microfluidic RT-qPCR, evaluated 20 published transcriptomic signatures, and used a feature selection algorithm to derive a novel treatment-monitoring signature which was evaluated in this cohort. We then evaluated this in an independent validation cohort of Indian children. Among 198 children (median 3.4 years, range 0.4-14.8), 37 (18.7%) had confirmed, 129 (65.2%) unconfirmed, and 32 (16.2%) unlikely TB. Baseline signature scores were the highest in children with confirmed TB and increased with radiographic severity. Scores declined within 2 weeks, with minimal change thereafter. Of the literature derived signatures, the Kaforou22 signature best differentiated baseline from EoT in confirmed TB (AUC 0.85, 95%CI 0.76-0.94) TB. We then discovered a novel 4-gene signature (FCGR1C, FLVCR2, LTF, and CDKN1C) that best differentiated baseline from EoT with an AUC of 0.89 (95%CI 0.81-0.97), with an AUC of 0.81 (95%CI 0.64-0.98) in an independent paediatric dataset. Blood transcriptomic signatures show promise as non-sputum tools for monitoring treatment response in children and may support biomarker-guided individualized TB treatment.</p>