Host response biomarkers of tuberculosis recurrence and treatment failure

BACKGROUND: Accurate detection of tuberculosis (TB) treatment failure and recurrence can improve disease control, but current sputum-based monitoring tools pose significant limitations. This study aimed to identify sputum-independent biomarkers for detecting and predicting TB treatment failure and recurrence. METHODS: Within the Pan-African TB Sequel study, we conducted a matched case-control study with 40 participants who had recurrent TB or treatment failure and 37 successfully treated controls matched by sex, age, and HIV status. Cases were classified as (a) non-converters with persistently positive sputum Mycobacterium tuberculosis (MTB) results during treatment, (b) reverters at the end of treatment (EOT), or (c) recurrence after EOT. Peripheral blood was collected at baseline, months 2, 4, 6, 9, and 12, and at suspected recurrence. MTB-specific T-cell activation markers (CD38, CD27, HLA-DR, Ki67) and transcriptomic signatures (Sweeney3, Risk6, MAMS6) were assessed and compared to the reference standard MTB culture and smear results. RESULTS: Here, we show that both MTB-specific T-cell activation and transcriptomic signatures detected non-conversion and TB recurrence at month 9 or 12 after treatment initiation. CD38 expression demonstrates 100% sensitive (95% CI: 56.6-100%) and 78% specific (95% CI: 56.5-99.4%) for detecting TB recurrence, with an AUC of 0.98 (95% CI: 91-100%). Among transcriptomic signatures, MAMS6, RISK6, and Sweeney3 achieve 75% sensitivity (95% CI: 50-100%) and 87-93% specificity (95% CI: MAMS6 0-100%, RISK6 0-93%, Sweeney3 0-100%), with comparable AUCs (0.78-0.83). Neither marker detected TB reversion at EOT. CONCLUSION: These sputum-independent biomarkers effectively identify TB disease, non-conversion and recurrence TB after EOT, whereas their utility in detecting TB reversion during treatment remains limited.